Direct Measurements: Bottle and Chamber Incubations

Supplemental MaterialStaehr et al. 2011

Electronic supplementary material

This document contains a detailed description and evaluations of the predominantly applied methodologies employed to measure or calculate gross primary production (Pg), respiration (R), and net ecosystem metabolism(Pn) in aquatic ecosystems.

Direct measurements: Bottle and chamber incubations

Of all the methods used to measure aquatic metabolism, perhaps the incubation of water in bottles and chambers has been the most widely used over time. Incubations of water in clear bottles exposed to varying irradiance levels for 1 to 24 hours have traditionally been used to determine plankton community photosynthesis (Gaarder and Gran 1927;Talling 1957;Kemp and Boynton 1980;Smith and Kemp 1995;Carignan et al. 1998;Gazeau et al. 2005) using both O2 and 14C-bicarbonate as tracers (Peterson 1980). In contrast, changes in O2 or TCO2 in dark bottles provide estimates of respiration. Rates are commonly extrapolated to represent 24-hour periods and if bottles are distributed in situ over the photic zone, depth and time integrated pelagic rates of Pg, Pn and R can be derived. Similar incubations of sediment chambers have been made to measure benthic photosynthesis and respiration and when combined with bottle incubations, it is possible to estimate Pg, Pn and R of entire ecosystems (e.g. Kemp et al. 1997; Gazeau et al. 2005).

Incubations are useful because they discern the metabolism of ecosystem components (i.e. pelagic vs. benthic), metabolic components (Pg, R and Pn), avoid the incorporation of terms for air-water gas exchange and water transport, and with current technologies, changes in O2 and TCO2 concentrations can be measured very accurately and precisely, allowing use of relatively short incubations (e.g. Reinthaler et al. 2006). Enclosing part of an aquatic system inside boundaries is attractive because it allows precise measurements with replication, permits experiments under changing conditions, and removes several uncertainties associated with other methods. Important disadvantages of these methods have generated interest in developing new approaches, despite the precision obtained with direct measurements.

Incubation experiments have two major problems (Bender et al. 1987): (1) processes that take place in bottles are not entirely equivalent to those that occur naturally, and (2) estimates of production and respiration rates from incubations often give ambiguous results and can fail to give complete community or ecosystem level rates. Respiration rates in the light can be several times higher than in the dark due to enhanced degradation of photosynthates produced in the light, photoenhancement of mitochondrial respiration, and photorespiration (Bender et al. 1987; Lewitus and Kana 1995). Accordingly, the assumption of unaltered respiration rates in light and darkness applied to calculate Pg may be erroneous. Perhaps a more critical problem is that to obtain ecosystem level estimates of planktonic and benthic metabolic rates integrated over large spatial and time scales, many individual measurements must be summed and averaged, thus leading to potentially large propagation of errors and failure to capture spatial heterogeneity (Sarma et al. 2005). In bottles and chamber, the exclusion of the grazing community and removal of the photosynthetic community from the mixing processes of the water column also make the measured rates problematic.

Open water methods: Diel changes in O2 or CO2

Open water methods (i.e. tracking changes in the products or raw materials of photosynthesis and respiration in situ to estimate rates) have been developed to avoid the limitations of bottle and chamber incubations. The diel O2 or TCO2/DIC method exploits the fact that net photosynthetic production of O2 /DIC occurs only during daylight whereas respiratory O2/DIC consumption occurs throughout the diel period but is the only metabolic process occurring at night. Thus Pg, R, and Pn can be directly quantified for a whole ecosystem by measuring changes in O2/DIC concentration throughout a 24-hour period and accounting for air-water gas exchange. This method has been used extensively in a variety of systems (e.g. Sargent and Austin 1949;;Smith and Key 1975;Cole and Fisher 1978;Kemp and Boynton 1980;Barnes 1983;Gattuso et al. 1993;D´Avanzo et al. 1996;Caffrey 2003;Staehr and Sand-Jensen 2007; Coloso et al. 2008).

The technique has several advantages. Diel O2 concentrations are easy to measure accurately and recent developments in remote sensors (e.g. YSI, Hydrolab, etc.) allow for measurements of O2 and physical parameters at frequencies of thirty minutes or less. Measurements of open water concentrations are generally thought to reflect the metabolism of the entire ecosystem, thus including all relevant components and avoiding the artefacts introduced by the use of bottles or chambers. Furthermore, O2, unlike inorganic C, has only a single chemical form in water; in contrast, dissolved inorganic C includes three chemical species which behave differently (CO2, HCO3-, and CO32-).

The disadvantages of this method are related to the assumptions needed to compute metabolic rates. First, an air-water exchange term must be measured or calculated in order to account for exchange of O2 with the atmosphere. Such estimates are highly variable in time and space and can be difficult to determine (Hartman and Hammond 1985;Kenney et al. 1988;Marino and Howarth 1993;Clark et al. 1996;Raymond and Cole 2001;Kremer et al. 2003). As a result of this complexity, air-water exchange coefficients must either be measured in each particular study (Borges et al. 2004) or estimated more crudely to allow multi-system comparisons (Caffrey 2004). Often a surrogate for turbulence at the air-water interface, such as wind speed, can be used (Wanninkhof 1992; Cole and Caraco 1998) which allows the research to at least approximate values for gas exchange. Second, conversion of metabolic rates measured as oxygen to units of carbon requires knowledge of the correct O2:C conversion factor and uncertainty may introduce error (Gazeau et al. 2005). Third, changes in O2 concentrations do not reflect anaerobic metabolism (e.g., sulphate, nitrate and iron reduction, methanogenesis) when waters become hypoxic or anoxic. Fourth, a single O2 sensor reflects some unknown horizontal and vertical zone of influence, and sensors in different locations within the same system can yield different estimates (Caraco and Cole 2002;Lauster et al 2006;Van de Bogert et al. 2007; Staehr et al. 2010). Lastly, diel changes in O2 are occasionally dominated by physical forces (air-water exchange, lateral/axial advection) that prevent establishment of sufficiently strong diel signal of O2 to compute metabolic rates.

The second and third disadvantages are avoided when the diel curve is based on measurements of diel changes in DIC (dissolved organic carbon) and exchange rates of CO2 between air and water, but the sensitivity of the DIC technique is usually much lower than that of O2 due to typically 5-20-fold higher concentrations of DIC than O2. Changes in pH (which change the proportions of the species within the inorganic C pool) and the precipitation or dissolution of CaCO3must be accounted for to avoid erroneous results and this complicates measurements and calculations (Bade and Cole 2006). Further, at low CO2 and high pH, the rate of atmospheric exchange of CO2 becomes dominated by the dissolution of CO3 at the air-water interface rather than by simple diffusion. This “chemically enhanced diffusion” can be much faster than simple diffusion (Emerson 1975) and difficult to estimate accurately (DeGrandpre et al. 1995; Wanninkhof and Knox 1996; Bade and Cole 2006).

The physical settings (e.g. size and depth) of a studied system determine whether vertical and horizontal heterogeneity becomes a problem in obtaining true whole-system metabolism. Applying multiple sondes at different depths and across the aquatic system increases the validity of the metabolic estimates should therefore be considered if whole-system estimates are required.

Open water methods: Response surface difference (RSD)

A method developed as an extension of the diel O2/TCO2 approach for tidal waters was developed by Swaney et al. (1999) and has been applied in other systems (e.g. Gazeau et al. 2005). Changes in O2/TCO2 may not represent local production and consumption due to advection and dispersion of non-local O2 pools due to tides, wind, and river flow. Salinity, however, can be used to trace the effects of mixing and flow on solute concentrations in tidal systems. Swaney et al. (1999) used depth profiles of O2 and salinity taken at several times during a day and predicted O2 concentration as a linear function of depth, salinity, and time, where salinity traces physical transport. The resulting rate of change of O2/TCO2 is used to compute Pg, R, and Pnin the fashion of the traditional diel approach, while accounting for air-water gas exchange.

Like the diel O2 approach, the RSD technique has several advantages. Oxygen concentrations are easy to measure accurately and frequently together with relevant physical parameters by sensors. Open water measurements reflect the metabolism of the entire ecosystem (all autotrophs and heterotrophs) and avoid bottle effects.

The disadvantages of this method are similarly linked to the assumptions needed for calculation of metabolic rates. An air-sea exchange term must be computed to account for exchanges of O2 or CO2 with the atmosphere (see above). Applying the correct O2:C conversion factor may introduce error (Gazeau et al. 2005) and changes in O2 concentration will not reflect anaerobic metabolism. Problems may also arise in stratified systems, where surface and bottom water may be nearly isolated, and non-biological or non-aerobic processes may be important. Lastly, diel changes in O2 are occasionally dominated by physical forces (air-sea exchange, lateral/axial advection) which may mask the biological diel O2 signal.

Open water methods: Oxygen isotopes

Oxygen isotopes have been used as alternatives to O2 concentrations to estimate, Pn, R, Pg:R, and Pg (Quay et al. 1995;Bender et al. 2000;Luz and Barkan 2000;Russ et al. 2004;Tobias et al. 2007). Rates of gross primary production integrated across broad spatial and temporal scales, can be estimated from the triple isotope (16O, 17O and 18O) composition of atmospheric and dissolved O2 and the rate of exchange of O2 between air and water (Luz et al. 1999;Sarma et al. 2005). Measurements of δ18O are also used to estimate P, R, and P:R separately over diel cycles in open waters (Tobias et al. 2007), in enclosed experiments (e.g. Bender et al. 2000;Luz et al. 2002), and where the P:R ratio at similar timescales can also be estimated using the ratio of 18O: 16O (Quay et al. 1995;Russ et al. 2004).

There are basically four approaches for using oxygen isotopes to evaluate metabolism. (1) The δ18O value alone can provide an estimate of the Pg:R ratio by the model of Quay et al. (1995). The key advantage is that no incubation is required and the Pg:R value reflects activity over the residence time of O2. (2) Incubation with 18O enriched water(Bender et al. 1987), which is a method that is summarized well by Ostrom et al. (2005), requires bottle incubations, but provides a direct estimate of Pg and, therefore, is less ambiguous than 14C. There isconcern, however, that in hypoxic environments there may be incomplete diffusion of the tracer into cells (Ostrom et al. 2005; Yacobi et al. 2007). (3) Diurnal variation in δ18O of O2 as described by Venkiteswaran et al. (2008), and finally, (4) the triple isotopes approach.

The triple isotope method, takes advantage of the fact that natural oxygen is composed of three stable isotopes: 16O, 17O and 18O, with different atomic abundances. The ratio of 18O:16O and 17O:16O in the atmosphere primarily depends on the isotopic composition of photosynthetically produced O2 from terrestrial and aquatic plants and on isotopic fractionation from respiration. The formation of a ∆17O anomaly in dissolved oxygen with respect to atmospheric air is discussed by Luz et al. (1999) and Luz and Barkan (2000). All biological and geochemical reactions on Earth produce a mass dependent relationship between the 17O:16O and 18O:16O of approximately 0.52. However, reactions with oxygen containing compounds, notably ozone, in the stratosphere with cosmic rays result in a mass independent relationship of approximately 1.0 (Luz and Barkan 2000). Some of the mass independent signal in ozone is transferred to stratospheric O2 which mixes with tropospheric O2 with the net result that there is anomalous abundance of 17O in tropospheric relative to the mass dependent relationship by approximately 0.3 per mil. During photosynthesis, O2 is produced by hydrolysis of the water molecule (H2O) with no fractionation such that the O2 produced has the same isotopic composition as the water. Further, all O2 produced by photosynthesis is mass dependent and thus has a distinct 17O abundance or anomaly relative to tropospheric O2. As O2 is produced by photosynthesis the 17O anomaly is reduced; the magnitude of the 17O shift is proportional to the rate of gas exchange. Therefore, knowledge of the gas exchange rate and 17O anomaly can be used to calculate the rate of Pg if the rate of air-sea exchange is known.

When the mixed layer is shallow and the thermocline is located in the photic zone, the calculated rates of Pg should be considered as minimum values, because some of the production takes place below the mixed layer. Conversely, when the deep mixing takes place in winter, the calculated Pg should be considered as a maximum value because some of the dissolved O2 with high ∆17O in the thermocline is incorporated into the mixed layer. Over an annual cycle, the winter excess should compensate for the summer deficit, and an annual integration of the calculated Pg is expected to reliably reflect the true integrated production in a given area.

The advantages of estimating Pg in the ecosystem with the triple oxygen approach is that Pg is integrated over time and space scales larger than incubation methods, making it possible to capture episodic production events. Although the 17∆ anomaly changed with time over 3 days at a given depth, the averaged mixed layer 17∆ anomaly was relatively constant, suggesting that 17∆ anomaly stores signals over the residence time of O2 (1 week in the case of Sagami Bay, Sarma et al. 2005).

There are a few disadvantages of the triple oxygen isotope technique. Three different processes, such as air-sea exchange, mixing with deep waters, and biological production or respiration can cause differences in δ17O and δ18O values in the mixed layer. The air-sea exchange rates are resource intensive and difficult to quantify. The Luz and Barkan (2000) model furthermore assumes that vertical mixing is negligible for changes in the 17∆ anomaly in the mixed water, which is not true if there is a subsurface chlorophyll maximum. Because the importance of mixing is difficult to evaluate, three dimensional modeling may be needed to fully evaluate its influence. Consistently, Pg estimated by use of the 17∆ anomaly is higher (up to 10 times) than that derived from O2 bottle incubations. Finally there is some debate about what value should be used for the mass dependent relationship (Luz and Barkan 2005).

Physical budgets

Pn has also been estimated at larger time and space scales by assuming Pn is equal to the residual concentrations of carbon (TOC or DIC), DIP (dissolved inorganic phosphorus), or O2 from a mass balance of physical inputs, outputs, and solute changes in pre-defined regions of a given system. Such budgets can be applied in any system where inputs and outputs are computed (Kemp et al. 1997), but in systems with gradients in a conservative tracer, such as salt, advective flow and diffusive mixing can be computed and coupled to TOC, DIC, DIP, or O2 data to calculate inputs and outputs, typically on a monthly time scale (see Smith et al. 1991; Karl and Lukas 1996; Gazeau et al. 2005; Testa and Kemp 2008 for complete descriptions of the mass balance calculations). A specific application of this approach is to estimate Pn by converting the net non-conservative DIP production to equivalent carbon units using a fixed C:DIP ratio (LOICZ; Smith et al. 1991; Gordon et al. 1996; Crossland et al. 2005).

The budget approach is advantageous because Pn is determined over large space and time scales and provides an integrated measure. Many different parameters can also be used for this analysis, and independent estimates of Pn obtained with the different parameters can be compared. Lastly, this approach is advantageous because rates of Pn are computed directly from straightforward and repeatable computations. An advantage of the ecosystem budget approach is that it allows computation of Pn from data commonly collected in conventional water quality monitoring programs, making it possible to compare metabolic processes among ecosystems (Gordon et al. 1996; Crossland et al. 2005).

The disadvantages of the budget approach are that many measurements or computations are required to estimate Pn. The fundamental limitation of this approach is that it estimates only Pn (not Pg, R, or P:R). The method also requires broad assumptions about process stoichiometry and metabolic pathways (Testa and Kemp 2008). Some of these rates may be difficult to quantify, may have large errors associated with them, or have been collected in different years. Data collected in different years may be affected by external forcing (e.g. varying freshwater inputs, temperature changes) and the differences among years may introduce error in the Pn estimate. As with many Pn methodologies, the mass balance approach also requires assumptions about C, N, and P ratios that may not always be appropriate. For DIC and O2, mass-balance air-sea gas exchange must be included.

Indirect approaches

Several other approaches have been applied to aquatic ecosystems to estimate Pn. Recent reports have exploited the disparate biological activity of O2 and N2 gas to estimate system metabolism at various depths (McNeil et al. 2006). Other investigators have evaluated Pn by measuring concentrations of O2 and CO2 in the water and atmosphere at one time during a day, and multiplying the difference between the measured concentration and the temperature and salinity corrected equilibrium concentration by an air-water exchange coefficient (Cole et al. 1994; del Giorgio et al. 1999; Cole et al. 2000; Sobek et al. 2005). Similar approaches have been applied in the open ocean (Najjar and Keeling 2000). Furthermore, monitoring the addition of inorganic C with special isotopic signatures (14C or 13C) offers an alternative approach to measure at least primary production. In the 1980’s 14C was added to several lakes in Canada, to determine how close bottle incubations would be to true whole-system metabolism (Hesslein et al. 1980). More recently inorganic 13C has been added to whole lakes (Cole et al. 2002; Pace et al. 2004; Pace et al. 2007) as a food-web tracer of autochthonously produced organic C. Coupled with models, these whole-system tracer additions also constrain the values of Pg and R, both at the ecosystem level and of specific components (Carpenter et al. 2005; Cole et al. 2006; Coloso et al. 2008).

References

Bade,D.L. and J.J.Cole 2006. Impact of chemically enhanced diffusion on dissolved inorganic carbon stable isotopes in a fertilized lake. Journal of Geophysical Research 111: C01014, doi:10.1029/2004JC002684.

Barnes,D.J. 1983. Profiling coral reef productivity and calcification using pH and oxygen electrodes. Journal of Experimental Mar ine Biology and Ecology 66: 149-161.