Mineralization of dissolved organic carbon in the ... - CiteSeerX
ELSEVIER
Marine Chemistry
51 (1995) 201-212
Mineralization of dissolved organic carbon in the Sargasso Sea Dennis A. Hansell, Nicholas R. Bates, Kjell Gundersen Bermuda Biological Station for Research, Inc., 17 Biological Lane, St. Georges. GE-01, Bermuda Received
14 December
1994; accepted
19 July 1995
Abstract The lability of dissolved organic carbon was estimated in the Sargasso Sea. Rates of DOC mineralization were estimated by monitoring, with high precision, the accumulation of CO, in dark incubation bottles after the removal of particles > 0.8 pm in size. The minimum incubation time used in these experiments was 24 h. Rates from three 24 h incubations conducted on water from 20 m fell in the narrow range of O-44-0.45 FM C d - ‘. These rates ranged from approximate equivalence to more than 100% greater than the concurrent rates of primary productivity, suggesting in some cases that gross primary productivity was underestimated or that there existed labile DOC produced earlier in time, thus supporting periods of net heterotrophy in the Sargasso Sea.
1. Introduction Dissolved organic matter (DOM) is the largest reservoir of organic carbon in the world’s oceans (Mackenzie, 1981). As such, changes in the size of the dissolved organic carbon (DOC) pool may be important in affecting atmospheric CO, levels on time scales of IOOO-10,000 yr (Hedges, 1992). The apparent relationship between DOC and atmospheric CO, has stimulated interest and controversy in the amount, production, export and mineralization of oceanic DOC over the last few years (e.g. Toggweiler, 1989; Bacastow and Maier-Reimer, 1991; Kirchman et al., 1991; Najjar et al., 1992). The size of the DOC pool in the surface layer of the open ocean reflects the balance between production and consumption. Seasonal variations in surface layer DOC concentrations have been demonstrated for both coastal and open ocean waters. Holmes et 03044203/95/$09.50 0 1995 Elsevier Science B.V. All rights reserved SSDI 0304-4203(95)00063-l
al. (1967) showed large accumulations of DOC that may have been tied to the occurrence of dinoflagellate blooms in La Jolla Bay, while Carlson et al. ( 1994) reported DOC accumulation during periods of water column stability in the Sargasso Sea. The trends shown in these papers are instructive because they demonstrate decoupling between the production and consumption terms. Trends do not, however, provide detailed information on the dynamics of the pool. To demonstrate an understanding of DOC dynamics, the processes comprising DOC production and mineralization must be quantified, and correlated to hydrographic or biologic variables. The gross production of DOC is difficult to measure because the numerous sources and time scales involved are complex and not easily resolved (Ignatiades and Fogg, 1973; Lancelot, 1979; Taylor et al., 1985; Eppley et al., 1981; Jumars et al., 1989; Ducklow and Carlson, 1992; Nagata and Kirchman,
202
D.A. Hansel1 et al./Marine
1992). DOC mineralization, to the contrary, occurs predominantly by the single process of oxidation by bacteria, so this term lends itself to direct quantification. Using controlled experimental designs, rates of DOC mineralization can be determined directly by measuring changes in concentrations of metabolic reactants (DOC and oxygen) or products (CO, and bacterial biomass). A traditional approach for determining DOC mineralization rates has been to monitor, in incubation bags or bottles, decreases in DOC or dissolved oxygen concentrations, and to compare these rates to changes in bacterial abundance and growth (e.g. Keys et al., 1935; Wakesman and Carey, 1935; Barber, 1968; Ogura, 1972; Kirchman et al., 1991; Amon and Benner, 1994). Quantification of mineralization via observation of covariant changes in DOC and bacterial abundance or growth is particularly difficult to apply in oligotrophic regions because of the low rates expected. In the Sargasso Sea, upper ocean DOC concentrations are typically 60-75 PM (Carlson et al., 19941, primary production rates are low (Lohrenz et al., 1992) and, consequently, mineralization rates are expected to be low. The primary difficulty is that the precision of the DOC measurement (e.g. Hansell, 1993) has been inadequate to resolve the changes expected on relevent time-scales. Thus another approach is warranted. Since the product of DOC mineralization is CO,, its production by microbial respiration can be used to quantify the mineralization process. Measurements of the increase in concentration of total carbon dioxide (TCO,) over various time courses of experimental incubations were made in this work using a high-precision gas extraction/coulometric method, similar to that described by Johnson et al. (1993). Since the incubations took place in the dark when photosynthetic uptake of CO, was negligible, CO, accumulation (due to microbial respiration) was assumed to be in molar equivalence to DOC mineralization by the microbes. This approach allowed determination of 24 h, 2 and 3 day mineralization rates, as well as the amount of DOC pool mineralization on the time scale of weeks. Finally, discussion is given on the possible sources of DOC that support the mineralization determined in the Sargasso Sea. Microbial utilization of semi-labile DOC, residual from the spring bloom period or from some other
Chemistry 51 (1995) 201-212
high DOC production period, may contribute to making the system net heterotrophic during periods of low primary production.
2. Methods 2.1. Sampling and experimental protocols Investigations of DOC mineralization were made with bottle incubations of water collected from 7 cruises on the R/V Weatherbird II between May 1993 and April 1994 at the sites of the U.S. JGOFS Bermuda Atlantic Time-series Study (BATS; 31”50’N, 64”lO’W) and Hydrostation “S” (32”10’N, 64’3O’W). Water was taken from 12 1 Niskin samplers, either from 20 m or from a suite of depths including 20, 70, 80, 200 and 1000 m. These sampling depths were chosen to approximate the depths of the shallow primary production maximum (upper 20 m), the particle maximum in the euphotic zone (70-100 m), and depths anticipated to exhibit little or no DOC production attributable to very recent primary production (200 and 1000 m). Collections of water for DOC experimentation were made during the mid- or late daylight period. Primary productivity measurements, against which the mineralization rates were compared, were determined on water from the same depths that had been collected at predawn of the same day, or the previous day. In both cases the location of the water column sampled was monitored by tracking a sediment trap array, which served to approximate horizontal water motion. Water was drawn from the Niskin samplers by gravity-filtration through an in-line 47 mm Nuclepore polycarbonate filter (0.8 pm pore size), which had been precleaned in 10% u/v HCl and well rinsed in Millipore Milli-Q water. Concurrent filtration of water from up to four Niskin samplers reduced filtration time to 1 h (filtering under vacuum was avoided to reduce the possibility of cell breakage). The filtration step was intended to reduce the abundance of bactivores and larger photoautotrophs, although no measurements were made of the abundance of particles larger than bacteria. Filtrate was collected into an acid-cleaned poly-
D.A. Hansel1 et al. /Marine
carbonate carboy and then dispensed into 0.5 or 1.0 1 Pyrex’ bottles, which had been acid-cleaned (with a 5% u/u HCl solution) and precombusted at 450°C. A small headspace was left in the bottle. The incubation bottles were fitted with ground glass stoppers and sealed with Apiezon’ L high vacuum grease to prevent degassing or ingassing of the sample. Care was taken to prevent the grease from coming into contact with the water (only the top of the stopper was greased). All incubation bottles were stored in the dark, with samples collected from I 200 m depth maintained at room temperature (N 20°C; within a few degrees of the temperature at the time of sampling) while deeper samples were refrigerated at 5°C. At the start of an incubation, and at various times up to 101 days, replicated control and incubated bottles (2-3 each) were terminated by adding 100 ~1 of saturated HgCl,. The stoppers were regreased, taped to provide positive closure of the bottle and stored in the dark until analysis for TCO,. At times, separate 1.0 1 Pyrex” bottles filled with filtrate from 20 m depth were incubated in order to monitor changes in bacterial abundance and growth. These bottles were not sealed with Apiezon” grease. DOC samples were collected from the grease-free bottles and stored in precombusted glass ampoules, which were subsequently flame sealed and frozen, or in Whatman EPA precleaned vials (40 ml), with teflon-lined caps, and frozen. Time points of the incubations were chosen with regard to the requirements of the experiment. Time courses of 0, 1, 2, and 3 days were chosen to investigate rapid changes in TCO, while longer time points of 6, 10 and 76 or 101 days were chosen to investigate long-term DOC mineralization. The poisoned time-zero samples served as controls for abiotic changes in TCO, during the experiments. These samples were maintained in the same storage conditions as the live samples and were analyzed at the end of the full incubation period. Time points between 0 and 1 days were not made due to the inability of the gas extraction-coulometric procedure to clearly distinguish < 0.3 PM changes in TCO, (see below). Throughout the text the terms labile, semi-labile and refractory DOC are used. The definitions used here are: labile DOC has turnover times of hours to days; semi-labile DOC has turnover times of weeks
Chemistry 51 (1995) 201-212
to months; refractory many years. 2.2. Geochemical
203
DOC has turnover
times
of
measurements
The introduction and development of the gas extraction/coulometric analytical technique for determining TCO, (Johnson et al., 1985,1987) has greatly improved our ability to follow metabolic changes in the production and consumption of TCO,. The precision of shipboard and laboratory measurements of TCO, can be better than 1 PM (Robinson and Williams, 1991; Johnson et al., 1993; Bates et al., 1995). A SOMMA (single-operator Multiparameter Metabolic Analyzer), similar to that described by Johnson et al. (1993), was used to control the pipetting and extraction of a seawater sample. The SOMMA was interfaced with a personal computer and coupled to a CO, coulometer detector (Model 5011; UIC Coulometrics Inc.). In the method, a calibrated volume of seawater was acidified with 8% u/u phosphoric acid, converting HCO; and CO:- to undissociated CO,. Free CO, was extracted by an inert carrier gas, such as ultra high purity nitrogen or helium gas, and absorbed by a coulometer cell solution containing ethanolamine, dimethylsulfoxide and a thymolphthalein indicator. Reaction of CO, with ethanolamine forms hydroxyethylcarbamic acid which is titrated with electrolytically generated hydroxide. The coulombs of electricity used to generate hydroxide are related to the moles of CO, absorbed by the solution using the Faraday constant (Johnson et al., 1993). The measurement was calibrated with known volumes of pure CO, (Wilke et al., 1993). This method offered highly precise and rapid (12- 15 min) sample analyses. The design of the SoMMA-coulometer system was made to facilitate the measurement of metabolic processes (Johnson et al., 1993). However, considerable variability in the day-to-day and cell-to-cell response of the coulometer solutions make it difficult to resolve changes in TCO, of less than 1 PM, especially under shipboard operation (SOMMA workshop, Seattle, USA. September 1993). Efforts were therefore made to improve the resolution of the extraction-coulometric technique (Bates, 1995; Bates et al., 1995). TCO, analyses were always conducted in the same climate controlled laboratory at the
D.A. Hansell et d/Marine
204
Bermuda Biological Station for Research and analyses of mineralization samples occured only when optimal operating conditions were met. TCO, analysis was typically made on the samples within a few weeks of the last bottle being terminated. Each time course of mineralization samples (e.g. 0, 1, 2, 3, 6, and 10 days) from one depth were analyzed on the same day using the same coulometer cell solutions. This procedure was followed in order to eliminate variability in TCO, analysis on different days due to slightly variable coulometer cell conditions. The coefficient of variation for TCO, analyses were typically better than 0.02% (0.4 /IM) for within- and between-bottle replicate analyses (for both mineralization experiments and water column measurements of TCO, at the BATS site>, a factor of 6 improvement over DOC analysis. The mean difference between duplicate analyses from the same bottle was approximately 0.015% (0.35 PM) for samples analyzed at the BATS site between 1992 and 1994 (Bates, 1995; Bates et al., 1995). Concentrations of DOC were determined by high-temperature combustion, similar to the method described by Hansel1 (1993) and Hansel1 et al. (1993). The quartz combustion tube (13 mm O.D.), maintained at 700°C contained 15 g Cuprox (Leeman Labs, Inc.), 15 g Sulfix (Wake Pure Chemical Industries, Inc.) and 3 g platinum pillows (Ionics, Inc.). Carbon dioxide was detected using a LICOR Model LI-6252 NDIR analyzer. The mean of the standard deviations from 36 individual determinations in this work (3-4 injections per determination) was 2.7 PM C. Calibrations were performed daily with a 4 point standard curve using glucose (maximum concentration of 100 PM C). The system blank (normally 7-8 PM C> was determined using the vialed Millipore Milli-Q water produced at the Bermuda Biological Station for Research. This water was cross referenced to the low-carbon reference water distributed by J. Sharp (University of Delaware) as part of the community DOC measurement intercalibration in 1994.
Chemistry 51 (1995) 201-212
Porter and Feig (1980). Bacterial growth was measured by the incorporation rate of methyl-tritiated thymidine (Fuhrrnan and Azam, 1982) using a conversion factor of 1.7 X 1018 bacterial cells per mole thymidine. Reference has been made to unpublished rates of primary productivity, measured by 14C uptake, and concentrations of particulate organic carbon (POC) measured in 1993 and 1994 that were determined by the BATS program. The methods used for all of the biological determinations are given in the BATS Methods Manual (Knap et al., 1993). 2.4. Tests for incubation
Tests were performed to evaluate the experimental design for artefacts. If the sampling technique introduced labile organics there should have been a noticable stimulation of the normally slow metabolic rates of bacteria at that depth. Incubations were performed twice on water collected from 1000 m and treated as described above (prefiltration). In the first experiment, the incubation was allowed to proceed for 6 days. A second experiment was run with water from 1000 m for 76 days as a check on the results of the first, short incubation. Also evaluated was the effect of filtration on bacterial abundance, bacterial growth, and DOC concentrations. 2.5. Test of precision To evaluate the limits of precision inherent to the experimental design, twelve incubation bottles were identically filled with water collected from 20 m (Hydrostation ‘ ‘S” site, June 1993), which had been prefiltered and dispensed as described above. Six of the bottles were “killed” immediately (described above) while the remainder were allowed to incubate for 5.4 days.
3. Results 3.1. Incubation
2.3. Biological
artefacts
artefacts
measurements
Bacterial abundance was determined by epifluorescence microscopy using 4’,6-diamidino-2-phenylindole (DAPI) as a stain according to the method of
Several tests were performed to evaluate the experimental design for artefacts, such as unintended stimulation or inhibition of bacterial respiration in the incubation bottles. Stimulation could occur due
D.A. Hansel1 et al. /Marine 201
I
04 . 0
I
2
.
I
I
I
I
4
6
8
10
Time
.
1
12
.
1
14
.
I
1
16
(days)
Fig. 1. Bacterial abundance (X 10’ 1.’ ) in an incubation container as a function of time in water collected from 20 m (May 1993) and filtered (0.8 pm pore size). One filter was counted per sample. The error bars are the standard deviation of 15 squares (each 2000 prn2) counted per filter.
to the introduction of labile organic carbon to the system (such as unintentional contamination with the Apiezon’ grease used to seal the bottles) or from the reduction of grazing pressure on the microbes. Inhibition of microbial metabolism could occur from physical damage to the organisms incurred with filtration or from the introduction of toxic compounds. Filtration of water collected from 20, 80 and 200 m (April 1994) did not result in a measurable increase in the DOC concentration (concentration stayed within f2 PM). Bacterial numbers were reduced by 11% with filtration (November 1993). In a separate test (May 1993; no concurrent TCO, production measurements were made), bacterial numbers doubled in the filtered samples over the first two days of the incubation (Fig. l), and decreased between days 6 and 16. Kirchman et al. (1991) observed similar bacterial abundance patterns, with a reduction of bacterial numbers after 2-3 days as the number of bactivore flagellates increased. In April 1994, initial bacterial growth rates in 20 m water was only 27% of growth in unfiltered seawater, but growth rates returned to near ambient levels in the course of a day (Fig. 2). Growth of bactivores in incubation chambers has been shown to occur after the first few days of incubation (Kirchman et al., 1991) but bactivore abundance was not monitored in this work. Because
Chemistry 51 (1995) 201-2 I2
205
of the anticipated lag in their appearance, the first few time points reported here should be good estimates of bacterial respiration. For the longer incubations (lo- 100 days), the presence of bactivores may improve the estimates of DOC utilization by bacteria. If marine bacteria have an assimilation efficiency of 10% (see below), then that percentage of DOC processed would be retained as bacterial biomass. Introduction of higher trophic levels such as bactivores should, in effect, increase mineralization of the DOC pool because they will respire a large fraction of the bacterial carbon ingested. TCO, production then becomes a strong indicator of total DOC utilization by bacteria over the long term. Finally, the assumption is made in this work that the respiration measured is not due to direct DOC utilization by microheterotrophs. If such a process is important in nature then the short-term rates reported may underestimate total DOC utilization because the organisms larger than 0.8 pm in size were intentionally removed. 3.2. Test of precision
of the protocol
This test was designed to quantify the variation in TCO, signal due to the filling of the bottles plus 1.0 7
t
0.8I d 5 % $ 3 s z ii
0.6-
0.4-
t 0.2-.
0.04 0
I
1
4
8
.
I
12
Time
I
16
I
20
I
24
28
(hours)
Fig. 2. Bacterial growth, measured as the incorporation rate of tritiated thymidine (pmole Tdr L-’ h”), in 0.8 pm filtered water (a). The water was taken from 20 m (April 1994). For comparison, the incorporation rate of unfiltered water at the start of the experiment is shown (0). Error bars are the standard deviations of the measurements (a = 3). The number of bacteria did not change significantly (P > 0.05; Student’s t-test; Sokal and Rohlf, 1969) over the time course of 1 day.
206
D.A. Hansel1 et d/Marine
Chemistry 51 (1992) 201-212
analytical limitations, as well as the variability imparted by the biological processes occurring over the period of incubation. The variability in the concentrations of TCO, in the six initial and final bottles represent the summed uncertainty associated with the protocol (Table 1). Initial bottles had a mean TCO, concentration of 2060.80 PM, with a standard deviation of 0.62 PM (CV = 0.030%), while the final bottles had a higher mean concentration of TCO, (2063.01 PM) and a larger standard deviation (1.13 ,uM; CV = 0.055%). The precision of the TCO, concentration determined for the initial bottles reflects the variability associated with filling the bottles and analytical procedures. The variability from the final bottles reflects the additional natural biological variability. The mean rate of TCO, production in these bottles was 0.41 PM dd ‘. 3.3. Short-term
0
2
3
4
Time (days) Fig. 3. Time courses of the production of TCO, (as proxy for the mineralization of DOC) during short-duration incubations. Error bars are the standard deviations of the measurements.
incubations
One-day incubations were performed to determine the initial rates of TCO, production resulting from the microbial mineralization of DOC. Three incubations were performed in the period from September 1993 until April 1994, each with water collected from 20 rn at the BATS or Hydrostation “S” sites. In the three experiments, the rate of TCO, production after 24 h was essentially identical at 0.440.45 FM d-’ (Fig. 3; Table 2). During the September time course the rate of TCO, production remained linear for the first three days of the incubation, while in all other cases the rates slowed beTable 1 Concentrations
1
tween days 1 and 3 (i.e. 0.35-0.45 PM d-’ at 2 days and 0.28-0.44 PM d-’ at 3 days). Incubations of intermediate duration (9 days) were performed with water collected from 3 depths in the upper 200 m of the water column during May 1993. The fastest initial rate of TCO, accumulation (3 day time point) was in the 20 m water, with rates at 70 and 200 m identical to one another (Fig. 4). The rates slowed after day 6, except at 70 m where TCO, production continued at a rate similar to the earlier time points. After 9 days, the amount of TCO, produced was similar at 20 and 70 m (1.36 and 1.57
of TCO? ( WM C) in the six initial and six final incubation
Initial
bottles. filled in June 1993
Final
Bottle
TCO,
Bottle
TCO,
1
2061.57
3 5 7 9 11
2061.35 2060.65 2060.23 2060.06 2060.83
(0.25) (0.00) (*I (0.31) (0.22) (0.12)
2 4 6 8 10 12
2064.19 2062.91 2064.06 2063.39 2061.56 2061.89
(0.43) (0.11) (0.47)
Grand Mean
2060.80
(0.62)
Grand Mean
2063.01
(1.13)
(0.48)
Final bottles were poisoned at 5.4 days. The concentration reported from each bottle is the mean of duplicate TCO, determinations (standard deviations in parentheses). Grand means and standard deviations were calculated from the eleven individual TCO, analyses performed in each treatment. * = Single determination from bottle.
D.A. Hansel1 et al./Marine
201
Chemistry 51 (1995) 201-212
Table 2 Rates of TCO, production ( PM C dd’ ; assumed equivalent to microbial respiration) and primary productivity (PP; PM C d- ’ ). All rates are from 20 m at the BATS site. Bacterial growth f PM C d- I >wascalculated using the conversion factor of 1.7X IO’* bacterial cehs per mole thymidine (Carlson et al., 199.5). A carbon conversion factor of 9.2 fg C (bacterial celI)- ’ was used (K. Gundersen, unpubl. data) Date September 1993 November 1993 April 1994
TCO,
PP
TCO, X PP-’
Bacterial growth
Efficiency
0.45 0.44 0.45
0.42 0.40 0.20
1.07 1.10 2.25
0.044 0.033 0.018
8.9 7.0 3.8
PM, respectively), while the amount produced at 200 m was 0.97 PM. Bacterial growth efficiencies were calculated from the bacteria1 growth rate, measured by thymidine uptake, and from the production of CO, (Table 2). Efficiencies ranged from 3.8 to 8.9%, with a mean of 6.2%. This range is in the low end of published efficiencies compiled by Jahnke and Craven (1995) and Carlson and Ducklow (1995), but similar to that estimated by Carlson and Ducklow (1995) for the BATS site. 3.4. Long-term
incubations
To determine the amount of DOC available to the microbes over periods of a few months, incubations were carried out for 76 or 101 days on water collected from 20, 80, 200 and 1000 m. The rate of DOC mineralization in samples from the mixed layer
was most rapid in the first 10 days of the incubation, and then slowed (Fig. 5). The size of the biologically available DOC pool (time scale of weeks to months) was variable, ranging from 3.3 to 6.6 FM C (Table 3) in the upper 200 m of the water column. DOC mineralization was not detectable in deep waters (1000 m) after 76 days. This last result indicates that inadvertent contamination of the samples prior to incubation can be avoided when caution is exercised. No clear correlation between the amount of semilabile DOC and depth or season was evident in the data presented in Table 3. As would be expected, the concentration of biologically available DOC at 1000 m was non-detectable. The DOC concentration at 1000 m, then, indicates the base concentration above which labile and semi-labile DOC would accumulate in the surface layer. The size of the labile plus semi-labile pool is ultimately defined by the difference in concentrations between the deep and surface layers.
I
-0.5 4
0
(%)
2
4
6
8
10
12 -I
Time (days) Fig. 4. Time courses of the production of TCO, (as proxy for the mineralization of DOC) during incubations of intermediate duration. Error bars are the standard deviations of the measurements.
20
40
60
80
100
120
Time (days) Fig. 5. Time courses of the production of TCO, (as proxy for the mineralization of DOC) during long-duration incubations.
D.A. Hansel1 et al. /Marine
208
Table 3 TCO, accumulation (PM) following incubations of water collected at the BATS and Hydrostation “S” (January 1994 only) sites, and the ambient DOC concentrations ( PM) Date
Depth(m)
TCO,
DOC
Duration (d)
November 1993 November 1993 November 1993 January 1994 January 1994
20 80 1000 20 200
4.25 3.31 nd 6.64 4.41
na na 46.2 b 69.6 = 72.6 a
76 76 76 101 101
Prefiltration was as described in the text. ’ Concentrations after 10 days of incubation. Initial samples had been contaminated during ampoulation. Values will underestimate the true concentrations by < 4 WM. b The mean of the concentrations of DOC at 1000 m (n = 5; standard deviation = 2.2 PM) determined from February-June, 1994. na = not available; nd = non-detectable.
4. Discussion The goals of the experiments were to determine the “instantaneous” rate of DOC mineralization by bacteria and to determine the quantity of biologically available DOC, on the time scale of weeks to months. To accurately determine the “instantaneous” rate, several requirements must be met: (1) the respiration measured must primarily be that of the organisms utilizing DOC as substrate; (2) there must be no stimulatory or inhibitory effects due to sample manipulation and; (3) the incubations have to be of short enough duration that the rates are representative of natural levels. Bacteria are the primary consumers of DOC in marine systems and they are the primary group of organisms that we expect to pass through the filters used. Keil and Kirchman (1991) found in the subarctic Pacific that 0.8 pm filtration passed 97% of the bacterial assemblage, but only 0.5% of the chlorophyll. This result suggests that most of the uptake of dissolved primary amines they measured was due to bacteria. In our work, 89% of the bacteria at 20 m passed through the filter. In the Sargasso Sea, bacterial biomass may make up a very large portion of the euphotic zone microbial carbon (Fuhrman et al., 1989), so passage of a fraction of the autotrophic assemblage should not have greatly impacted the rates we measured. We were unable to resolve an increase in the
Chemistry 51 (1995) 201-212
concentrations of DOC due to filtration, but studies using more sensitive techniques have shown that some DOC should be expected to be released from particles during filtration (Goldman and Dennett, 1985). Keil and Kirchman (1991), performing similar gravity filtrations using 0.8 pm polycarbonate filters, found that dissolved free amino acid concentrations increased by an average of 11%. Hilmer and Bate (1989) reported a total photosynthate release of only 3%, this value apparently including naturally released, 14C-labeled DOC. Goldman and Dennett (1985) and Hilmer and Bate (1989) focused on recently fixed carbon, and not on the total mass of particulate organic carbon (POC) released as DOC during filtration. The magnitude of the total release is likely a function of the POC load, which in the Sargasso Sea is low (Michaels et al., 1994). During the period when water was collected for the experiments reported here, the ambient concentrations of POC at 20 m had a narrow range of 1.6-2.0 PM C (or < 3% of the DOC concentration). If it is assumed that 5% of the POC pool (exclusive of bacteria) can be passed through a filter and made available as DOC, then the increase to the ambient DOC pool was no more than 0.08-o. 1 PM C. This level of increase must be viewed as a “trace” addition to the biologically available DOC pool, which is shown below to be an order of magnitude or more larger in the Sargasso Sea. If a stimulation of the bacterial population occured at this level, it was below the level of sensitivity available for resolving its contribution. Rationalization on why the sampling procedure was not thought to stimulate bacterial metabolism has been presented. It appears, though, that for several hours immediately following filtration, bacterial growth rates may be reduced (Fig. 2). Growth rates returned to levels near the initial, unfiltered value within 25 h of the start of the incubation. The reductions in growth rate were contrary to the results of Coffin et al. (1993) who found the highest rates of microbial respiration during the hours immediately following filtration. Studies on bacterial respiration and growth have suggested that bacteria are rapidly respiring a very small pool of highly labile, recently produced DOC (Fuhrman, 1987). Coffin et al. (1993), for example, working in near-shore waters, showed a reduction in the rate of oxygen consumption 5- 12 h
D.A. Hansel1 et al./Marine
after 1.0 pm filtration, apparently due to exhaustion of the most labile fraction of DOC. Application of the method reported here in oligotrophic waters cannot resolve changes over short time scales, leaving open the possibility that the rates based on 24 h incubations are underestimates. In order to better assess the validity of the rates of DOC mineralization we report, the rates can be compared to the few other reports of microbial respiration or DOC consumption in the Sargasso Sea. Kepkay et al. (1990) measured respiration in water collected from 10 m and prefiltered through a 2 pm filter. Mean oxygen consumption was 0.37 /_LM 0, dd’ (or 0.27 PM C d-‘, using their molar respiratory quotient of 1.35) while primary productivity was 0.38 PM dd’. The rate of DOC mineralization they reported was 40% lower than our value at 20 m. The mean of the whole water (20-25 m) respiration rates reported by Williams and Jenkinson (1982) was 1.24 PM C d-‘, or 2.7 times the rate we measured. Bumey (1986) determined the uptake of dissolved carbohydrate at 30 m to be 0.12 PM C h- ’ while Carlson and Ducklow (1995) reported a range of 0.04-0.10 PM C hh’ of DOC utilization at 10 m. Much of the labile plus semi-labile DOC in the surface layer of the ocean is likely composed of carbohydrates (Benner et al., 1992) so Bumey’s results should be directly comparable to the bulk measurements of DOC utilization. An unexpected result of our work was that the rates of DOC mineralization were high relative to concurrent rates of primary productivity (Table 2). If the mineralization of DOC is indeed tightly coupled to its production, then it follows that the rate of microbial respiration should not ordinarily be equal to or greater than the gross rate of primary productivity. This necessity stems from other loss terms expected, such as the grazing of the daily primary production in the Sargasso Sea that must take place. Roman et al. (1993) reported grazing rates in August of 0.6 mg C mP2 h- ’ for mesozooplankton. No published data have been found for microzooplankton grazing on phytoplankton in the Sargasso Sea but their impact is likely significant. So, in order for near-balance between the measured rates of DOC consumption and measured primary productivity to exist then: (1) there must exist significant pools of semi-labile DOC, the utilization of which pushes the
Chemistry 51 (1995) 201-212
209
system into long periods of net heterotrophy, or; (2) the measured primary productivity underestimated gross primary productivity at the time of the experiments. 4.1. Sources of semi-labile
DOC
Two source terms for the semi-labile DOC pool, independent of recently produced photosynthate, can be identified. These are the photo-oxidation of refractory DOC, of a deep-water source, in the upper few meters of the water column and the utilization of DOC produced during the previous spring bloom or some other period of high DOC production, which is semi-refractory to microbial utilization and slowly mineralized. Recent work has demonstrated that the photo-oxidation of biologically refractory DOC, mixed to the surface during winter overturn, may be an important sink for deep-water DOC (Kieber et al., 1989; Mopper et al., 1991). This deep-water pool of DOC has been characterized as refractory to microbial consumption (Barber, 1968; Table 3). Mopper et al. (1991), however, measured the production of low molecular weight carbonyl compounds in the surface waters of the Sargasso Sea (4.1 nM C h-‘, exclusive of CO production they measured). They implicated natural levels of UV light as the causative agent and suggested deep-water DOC as the source. If this rate is doubled to account for other low molecular weight organics not measured (e.g. organic acids), and 8 h dd’ of adequate light is assumed, then the daily rate of production of low molecular weight, labile DOC may be 0.07 PM d.- ‘. The rate of production should be equal to its utilization by bacteria. The extent to which this process occurs at 20 m, the shallowest depth examined in our work, is unknown. The second source of semi-labile DOC is that produced and accumulated during the Sargasso Sea spring bloom or during other periods of high DOC production. The region represented by the BATS site is characterized by a strong seasonality in mixed layer depths, a winter recharge of surface nutrients and a vernal phytoplankton bloom (Lohrenz et al., 1992; Michaels et al., 1994). The 1993 spring bloom, the bloom preceding our experiments, reached levels of primary production as high as 880 mg C mm 2 d-‘. Unlike the clear spring bloom in 1993, there was no major bloom in 1994, apparently due to the
210
D.A. Hansel1 et al. / Marine Chemistry 51 (1995) 201-212
absence of wintertime convective overturn of the water column. Rates of primary production remained below 400 mg C mm2 dd’ throughout the 1994 winter/spring period. The “short-term” rates that we report in Fig. 3 were determined in fall, following the 1993 bloom, and the winter and spring of 1994. Carlson et al. (1994), reporting seasonal variability in the concentrations of DOC at the BATS site, found an accumulation of DOC above background concentrations (0.6-0.8 mol C mP2 above late summer values) during two consecutive spring blooms. The concentrations of DOC decreased following the bloom periods, in both the upper water column (O-100 m) and in the layer immediately below (loo-250 m). Decreases in DOC concentrations in both layers may approximate net rates of mineralization (changes in concentration due to advection could not be resolved). The timing of DOC accumulation in the deep layer ties its production to the spring bloom. Strong seasonal stratification following the spring bloom effectively seals the deeper layer, largely eliminating additional inputs of DOC from the euphotic zone by mixing. The consumption of DOC in the lower layer after stratification should represent the mineralization of semi-refractory DOC produced during the spring bloom. The mean rate of decrease in the DOC concentration (loo-250 m) from March to April 1993 in the Carlson et al. (1994) data was 0.17 PM dd ’ . The rate of DOC disappearance decreased thereafter. This finding is in reasonable agreement with the DOC mineralization rate of 0.07 PM dd’ measured in May, 1993 at 200 m (Fig. 4). We suggest that DOC mineralization occuring below the seasonal thermocline may largely be supported by substrate produced during the previous spring bloom. If due to the spring bloom, then the rate of mineralization below the seasonal thermocline should be indicative of the microbial consumption of the same fraction of DOC (residual from the spring bloom) in the surface mixed layer. For budgeting purposes at 20 m, the rate of DOC mineralization in the deep layer (at 200 m) is taken to represent the rate of mineralization of the same fraction of semi-refractory DOC residing in the euphotic zone. That rate is 0.07 PM d-l, or 15% of the total mineralization rate (0.45 FM C d-l) at 20 m. In
balance, 85% of the daily rate of DOC mineralization rate must be supported by recent primary production (the contribution of photo-oxidation at 20 m is assumed to be negligible). This result requires that 0.38 PM C d-’ (or 90%) of the daily primary production (0.42 PM dd ’ ; Table 2) measured in September 1993, for example, be consumed by bacteria. This value is somewhat greater than that of Biddanda et al. (1994), who found that 69% of the daily primary production in the slope waters of the northern Gulf of Mexico was processed by the bacterioplankton. 4.2. Underestimation
of gross primary productiuity
We have no evidence that there is a significant difference between net primary production measured in the BATS program (here defined as the 14C activity in filtered particles, corrected with dark incubations, after 24 h, single-time point incubations) and gross primary production (here defined as summed fixation of carbon into the particulate and dissolved organic pools). There remains, though, a lingering concern that in many areas the 14C method for primary production underestimates the true production by large amounts (Banse, 1994). If gross primary production is indeed underestimated then much of the underestimate might be found in the dissolved organic pool. Lancelot (1979) found 14C to accumulate into the DOC pool at rates approximately l/3 of the accumulation in particles, while Laws et al. (1984) found accumulation of labelled DOC ranged from non-detectable to 18% of the particulate pool after 24 h incubations. The transfer of label from the particulate to the dissolved pool these studies report can be large, but they are not on the order of equivalence between net production and DOC consumption as required by the results of this work. The few rates of DOC consumption reported for the Sargasso Sea (listed above) are, like those of this work, high relative to anticipated concurrent levels of primary productivity (rates at the BATS site range from 2 to 20 mg C me3 dd ‘; BATS Data Reports). Note, though, that the 14C method of determining DOC production traces the transfer of recent photosynthate to solution, while measurements of microbial utilization of DOC encompass all DOC source terms including, but not limited to, recent photosyn-
D.A. Hansel1 et al./Marine
thate. So the discrepancy that exists (i.e. the ratio of 14C estimated DOC production to primary productivity, as reported in the literature, is generally smaller than the expected ratios of DOC mineralization to primary productivity in the Sargasso Sea) may be due to processes that contribute to the production of labile or semi-labile DOC but that cannot be classified as loss of recent photosynthate. Some processes have been hypothesized (sloppy feeding, loss of DOC from egested fecal pellets, dissolution of sinking particles, viral disruption of bacteria) but none have been adequately quantified. These secondary processes of DOC production clearly need to be evaluated for their role in supporting bacterial growth.
Acknowledgements We thank Drs. A.F. Michaels and A.H. Knap for access to the BATS data from 1993 and 1994. Support for DAH came from NSF grants OCE9311012 and OPP-9317200, while support for NRB and KG came from NSF grant OCE-9301950. J.1 Hedges, C.A. Carlson, E.T. Peltzer, R.G. Keil and one anonymous reviewer are thanked for their helpful and constructive comments. The US Department of Energy CO, Science Team is acknowledged for the loan of a SoMMA-Coulometer TCO, Analyzer to BBSR. This is Contribution No. 1404 from the Bermuda Biological Station for Research, Inc.
References Amon, R.W. and Bemier, R., 1994. Rapid cycling of high-molecular weight dissolved organic matter in the ocean. Nature, 369: 549-552. Bacastow, R. and Maier-Reimer, E., 1991. Dissolved organic carbon in modeling oceanic new production. Global Biogeochem. Cycles, 5: 71-85. Banse, K., 1994. Commentary: An unpleasant note about the carbon-14 method of estimating plankton photosynthesis. US JGOFS Newslett., 6: 3-15. Barber, R.T., 1968. Dissolved organic carbon from deep water resists microbial oxidation. Nature, 220: 274-275. Bates, N.R., 1995. Investigation of the Physical and Biological Controls of the Oceanic CO, System in the Sargasso Sea. Ph.D. Dissertation. Univ. Southampton, Southampton, 277 pp. Bates, N.R., Michaels, A.F. and Knap, A.H., 1995. Seasonal and interannual variability of the oceanic carbon dioxide system at
Chemistry 51 (1995~ 201-212
211
the U.S. JGOFS Bermuda Atlantic Time-Series Site. Deep-Sea Res. 11, in press. Benner, R., Pakuiski, J.D., McCarthy, M., Hedges, J-1. and Hatcher, P.G., 1992. Bulk chemical characteristics of dissolved organic matter in the ocean. Science, 255: 1561- 1564. Biddanda, B., Opsahl, S. and BeMer, R., 1994. Plankton respiration and carbon flux through bactetioplankton on the Louisiana shelf. Limnol. Oceanogr., 39: 1259-1275. Bumey, C.M.. 1986. Bacterial utilization of total in situ dissolved carbohydrate in offshore waters. Limnol. Oceanogr., 31: 427431. Carlson, C.A. and Ducklow, H.W., 1995. Bacterial growth and dissolved organic carbon utilization in the oligotrophic Sargasso Sea. Aquat. Microb. Ecol., submitted. Carlson, CA., Ducklow, H.W. and A.F. Michaels, 1994. Annual flux of dissolved organic carbon from the euphotic zone in the northwestern Sargasso Sea. Nature, 371: 405-408. Carlson, C.A., Ducklow, H.W., Sleeter, T.D., 1995. Stocks and dynamics of bactetioplankton in the northwestern Sargasso Sea. Deep-Sea Res., in press. Coffin, R.B., Connolly, J.P. and Harris, P.S., 1993. Availability of dissolved organic carbon to bactetioplankton examined by oxygen utilization. Mar. Ecol. hog. Ser., 101: 9-22. Ducklow, H.W. and Carlson, C.A., 1992. Oceanic bacterial production. Adv. Microb. Ecol., 12: 113-181. Eppley, R.W., Honigan, S.G., Fuhrman, J.A., Brooks. E.R., Price, C.C. and Sellner, K., 1981. Origins of dissolved organic matter in Southern California coastal waters: experiments on the role of zooplankton. Mar. Ecol. Prog. Ser., 6: 149-159. Fuhrman, J.A. and Azam, F., 1982. Thymidine incorporation as a measure of heterotrophic bacterial production in marine surface waters: evaluation and field results. Mar. Biol., 66: 109-120. Fuhrman, J.A., 1987. Close coupling between release and uptake of dissolved free amino acids in seawater studied by an isotope dilution technique. Mar. Ecol. Prog. Ser., 37: 45-52. Fuhrman, J.A., Sleeter, T.D., Carlson, C.A. and Proctor, L.M., 1989. Dominance of bacterial biomass in the Sargasso Sea and its ecological implications. Mar. Ecol. Prog. Ser., 57: 207-217. Goldman, J.C. and Dennett, M.R., 1985. Susceptibility of some marine phytoplankton species to cell breakage during filtration and post-filtration rinsing. J. Exp. Mar. Biol. Ecol., 86: 47-58. Hansell, D.A., 1993. Results and observations from the measurement of DOC and DON using high temperature catalytic combustion techniques. Mar. Chem., 41: 195-202. Hansell, D.A., Williams, P.M. and Ward, B.B., 1993. Measurements of DOC and DON in the Southern California Bight using oxidation by high temperature combustion. Deep-Sea Res., 40: 219-234. Hedges, J.I., 1992. Global biogeochemical cycles: Progress and problems. Mar. Chem., 39: 67-93. Hilmer, T. and Bate, G.C., 1989. Filter types, filtration and post-filtration treatment in phytoplankton production studies. J. Plankt. Res., 11: 49-63. Holmes, R., Williams, P.M. and Eppley, R.W., 1967. Red water in La Jolla Bay. Limnol. Oceanogr., 12: 5033512. Ignatiades, L. and Fogg, G.E., 1973. Studies on the factors
212
D.A. Hansel1 et al./Marine
affecting the release of organic matter by Skeleronema costarum (Greville) Cleve in culture. J. Mar. Biol. Assoc. UK, 53: 937-956. Jahnke, R.A. and Craven, D.B., 1995. Quantifying the role of heterotrophic bacteria in the carbon cycle: A need for respiration rate measurements. Limnol. Oceanogr., 40: 436-441. Johnson, K.M., King, A.E. and Sieburth, J.McN., 1985. Coulometric TCO, analyses for marine studies: an introduction. Mar. Chem., 16: 61-82. Johnson, K.M., Sieburth, J.M., Williams, P.J. 1eB. and BrandStrom, L., 1987. Coulometric total carbon dioxide analysis for marine studies: automation and calibration. Mar. Chem., 21: 117-133. Johnson, K.M., Wills, K.D., Butler, D., Johnson, W.K. and Wong, C.S., 1993. Coulometric total carbon dioxide analysis for marine studies: maximizing the performance of an automated gas extraction system and coulometric detector. Mar. Chem., 44: 167-188. Jumars, P.A., Penry, D.L., Baross, J.A., Perry, M.J. and Frost, B.W., 1989. Closing the microbial loop: Dissolved carbon pathway to heterotrophic bacteria from incomplete ingestion, digestion and absorption in animals. Deep-Sea Res., 36: 483495. Knap, A.H., Michaels, A.F., Dow, R.L., Johnson, R.J., Gundersen, K., Sorenson, J.C., Close, A.R., Howse, F.A., Hammer, M., Bates, N.R., Doyle, A. and Waterhouse, T., 1993. BATS Methods Manual. US JGOFS Planning Off., Woods Hole, MA. Keil, R.G. and Kirchman, D.L., 1991. Contribution of dissolved free amino acids and ammonium to the nitrogen requirements of heterotrophic bacterioplankton. Mar. Ecol. Prog. Ser., 73: l-10. Kepkay, P.E., Harrison, W.G. and Irwin, B., 1990. Surface coagulation, microbial respiration and primary production in the Sargasso Sea. Deep-Sea Res., 37: 145-155. Keys, A., Christensen, E.H. and Krogh, A., 1935. The organic metabolism of sea-water with special reference to the ultimate food cycle in the sea. J. Mar. Biol. Assoc. UK, 20: 181-196. Kieber, D.J., McDaniel, J. and Mopper, K, 1989. Photochemical source of biological substrates in sea water: implications for carbon cycling. Nature, 341: 637-639. Kirchman, D.L., Suzuki, Y., Garside, C. and Ducklow, H.W., 1991. High turnover rates of dissolved organic carbon during a spring phytoplankton bloom. Nature, 352: 612-614. Lancelot, C., 1979. Gross excretion rate of natural marine phytoplankton and heterotrophic uptake of excreted products in the Southern North Sea, as determined by short-term kinetics. Mar. Ecol. Prog. Ser., 1: 179-186. Laws, E.A., Redalje, D.G., Haas, L.W., Bienfang, P.K., Eppley, R.W., Harrison, W.G., Karl, D.M. and Marra, J., 1984. High phytoplankton growth and production rates in oligotrophic Hawaiian coastal waters. Limnol. Oceanogr., 29: 943-948. Lohrenz, S.E., Knauer, G.A., Asper, V.L., Tuel, M., Michaels, A.F. and Knap, A.H., 1992. Seasonal variability in primary production and particle flux in the northwestern Sargasso Sea:
Chemistry S1 (1995) 201-212 U.S. JGOFS Bermuda Atlantic Time-series Study. Deep-Sea Res., 39: 1373-1391. Mackenzie, F.T., 1981. Global carbon cycle: Some minor sinks for CO,. In: G.E. Likens, F.T. Mackenzie, J.E. Richey, J.R. Sedell and K.K. Turekian (Editors), Flux of Organic Carbon by Rivers to the Oceans. US Dep. Energy, Washington, DC, pp. 360-384. Michaels, A.F., Knap, A.H., Dow, R.L., Gundersen, K., Johnson, R.J., Sorensen, J., Close, A., Knauer, G.A., Lohrenz, SE., Asper, V.A., Tuel, M. and Bidigare, R., 1994. Seasonal patterns of biogeochemistry at the U.S. JGOFS Bermuda Atlantic Time-series Study site. Deep-Sea Res., 41: 1013-1038. Mopper, K., Zhou, X., Kieber, R.J., Kieber, D.J., Sikorski, R.J. and Jones, R.D, 1991. Photochemical ‘degradation of dissolved organic carbon and its impact on the oceanic carbon cycle. Nature, 353: 60-62. Nagata, T. and Kirchman, D.L., 1992. Release of macromolecular organic complexes by heterotrophic marine flagellates. Mar. Ecol. Prog. Ser., 83: 233-240. Najjar, R.G., Sarmiento, J.L. and Toggweiler, J.R., 1992. Downward transport and fate of organic matter in the ocean: simulations with a general circulation model. Global Biogeochem. Cycles, 6: 45-76. Ogura, N., 1972. Rate and extent of decomposition of dissolved organic matter in surface seawater. Mar. Biol., 13: 89-93. Porter, K.G. and Feig, Y.S., 1980. The use of DAPI for identifying and counting aquatic microflora. Limnol. Oceanogr., 25: 943-948. Robinson, C. and Williams, P.J. leB., 1991. Development and assessment of an analytical system for the accurate and continual measurement of total dissolved inorganic carbon. Mar. Chem., 34: 157-175. Roman, M.R., Dam, H.G., Gauzens, A.L. and Napp, J.M., 1993. Zooplankton biomass and grazing at the JGOFS Sargasso Sea time series station. Deep-Sea Res., 40: 883-901. Sokal, R.R. and Rohlf, F.J., 1969. Biometry. Freeman, New York, NY, 859 pp. Taylor, G.T., Iturriaga, R. and Sullivan, C.W., 1985. Interactions of bactiverous grazers and heterotrophic bacteria with dissolved organic matter. Mar. Ecol. Prog. Ser., 23: 129-141. Toggweiler, J.R., 1989. Is the downward flux of dissolved organic matter (DOM) important in carbon transport? In: W.H. Berger, V.S. Smetacek and G. Wefer (Editors), Productivity of the Ocean: Present and Past. Wiley, Chichester, pp. 65-83. Wakesman, S.A. and Carey, C.L., 1935. Decomposition of organic matter in sea water by bacteria. II. Influence of addition of organic substances upon bacterial activities. J. Bacterial., 29: 545-561. Wilke, R., Wallace, D.W.R. and Johnson, K.M., 1993. A waterbased, gravimetric method for the determination of gas sampling loop volume. Anal. Chem., 65: 2403-2406. Williams, P.J. 1eB. and Jenkinson, N.W., 1982. A transportable microprocessor-controlled precise Winkler titration suitable for field station and shipboard use. Limnol. Oceanogr., 27: 576-584.
Marine Chemistry
51 (1995) 201-212
Mineralization of dissolved organic carbon in the Sargasso Sea Dennis A. Hansell, Nicholas R. Bates, Kjell Gundersen Bermuda Biological Station for Research, Inc., 17 Biological Lane, St. Georges. GE-01, Bermuda Received
14 December
1994; accepted
19 July 1995
Abstract The lability of dissolved organic carbon was estimated in the Sargasso Sea. Rates of DOC mineralization were estimated by monitoring, with high precision, the accumulation of CO, in dark incubation bottles after the removal of particles > 0.8 pm in size. The minimum incubation time used in these experiments was 24 h. Rates from three 24 h incubations conducted on water from 20 m fell in the narrow range of O-44-0.45 FM C d - ‘. These rates ranged from approximate equivalence to more than 100% greater than the concurrent rates of primary productivity, suggesting in some cases that gross primary productivity was underestimated or that there existed labile DOC produced earlier in time, thus supporting periods of net heterotrophy in the Sargasso Sea.
1. Introduction Dissolved organic matter (DOM) is the largest reservoir of organic carbon in the world’s oceans (Mackenzie, 1981). As such, changes in the size of the dissolved organic carbon (DOC) pool may be important in affecting atmospheric CO, levels on time scales of IOOO-10,000 yr (Hedges, 1992). The apparent relationship between DOC and atmospheric CO, has stimulated interest and controversy in the amount, production, export and mineralization of oceanic DOC over the last few years (e.g. Toggweiler, 1989; Bacastow and Maier-Reimer, 1991; Kirchman et al., 1991; Najjar et al., 1992). The size of the DOC pool in the surface layer of the open ocean reflects the balance between production and consumption. Seasonal variations in surface layer DOC concentrations have been demonstrated for both coastal and open ocean waters. Holmes et 03044203/95/$09.50 0 1995 Elsevier Science B.V. All rights reserved SSDI 0304-4203(95)00063-l
al. (1967) showed large accumulations of DOC that may have been tied to the occurrence of dinoflagellate blooms in La Jolla Bay, while Carlson et al. ( 1994) reported DOC accumulation during periods of water column stability in the Sargasso Sea. The trends shown in these papers are instructive because they demonstrate decoupling between the production and consumption terms. Trends do not, however, provide detailed information on the dynamics of the pool. To demonstrate an understanding of DOC dynamics, the processes comprising DOC production and mineralization must be quantified, and correlated to hydrographic or biologic variables. The gross production of DOC is difficult to measure because the numerous sources and time scales involved are complex and not easily resolved (Ignatiades and Fogg, 1973; Lancelot, 1979; Taylor et al., 1985; Eppley et al., 1981; Jumars et al., 1989; Ducklow and Carlson, 1992; Nagata and Kirchman,
202
D.A. Hansel1 et al./Marine
1992). DOC mineralization, to the contrary, occurs predominantly by the single process of oxidation by bacteria, so this term lends itself to direct quantification. Using controlled experimental designs, rates of DOC mineralization can be determined directly by measuring changes in concentrations of metabolic reactants (DOC and oxygen) or products (CO, and bacterial biomass). A traditional approach for determining DOC mineralization rates has been to monitor, in incubation bags or bottles, decreases in DOC or dissolved oxygen concentrations, and to compare these rates to changes in bacterial abundance and growth (e.g. Keys et al., 1935; Wakesman and Carey, 1935; Barber, 1968; Ogura, 1972; Kirchman et al., 1991; Amon and Benner, 1994). Quantification of mineralization via observation of covariant changes in DOC and bacterial abundance or growth is particularly difficult to apply in oligotrophic regions because of the low rates expected. In the Sargasso Sea, upper ocean DOC concentrations are typically 60-75 PM (Carlson et al., 19941, primary production rates are low (Lohrenz et al., 1992) and, consequently, mineralization rates are expected to be low. The primary difficulty is that the precision of the DOC measurement (e.g. Hansell, 1993) has been inadequate to resolve the changes expected on relevent time-scales. Thus another approach is warranted. Since the product of DOC mineralization is CO,, its production by microbial respiration can be used to quantify the mineralization process. Measurements of the increase in concentration of total carbon dioxide (TCO,) over various time courses of experimental incubations were made in this work using a high-precision gas extraction/coulometric method, similar to that described by Johnson et al. (1993). Since the incubations took place in the dark when photosynthetic uptake of CO, was negligible, CO, accumulation (due to microbial respiration) was assumed to be in molar equivalence to DOC mineralization by the microbes. This approach allowed determination of 24 h, 2 and 3 day mineralization rates, as well as the amount of DOC pool mineralization on the time scale of weeks. Finally, discussion is given on the possible sources of DOC that support the mineralization determined in the Sargasso Sea. Microbial utilization of semi-labile DOC, residual from the spring bloom period or from some other
Chemistry 51 (1995) 201-212
high DOC production period, may contribute to making the system net heterotrophic during periods of low primary production.
2. Methods 2.1. Sampling and experimental protocols Investigations of DOC mineralization were made with bottle incubations of water collected from 7 cruises on the R/V Weatherbird II between May 1993 and April 1994 at the sites of the U.S. JGOFS Bermuda Atlantic Time-series Study (BATS; 31”50’N, 64”lO’W) and Hydrostation “S” (32”10’N, 64’3O’W). Water was taken from 12 1 Niskin samplers, either from 20 m or from a suite of depths including 20, 70, 80, 200 and 1000 m. These sampling depths were chosen to approximate the depths of the shallow primary production maximum (upper 20 m), the particle maximum in the euphotic zone (70-100 m), and depths anticipated to exhibit little or no DOC production attributable to very recent primary production (200 and 1000 m). Collections of water for DOC experimentation were made during the mid- or late daylight period. Primary productivity measurements, against which the mineralization rates were compared, were determined on water from the same depths that had been collected at predawn of the same day, or the previous day. In both cases the location of the water column sampled was monitored by tracking a sediment trap array, which served to approximate horizontal water motion. Water was drawn from the Niskin samplers by gravity-filtration through an in-line 47 mm Nuclepore polycarbonate filter (0.8 pm pore size), which had been precleaned in 10% u/v HCl and well rinsed in Millipore Milli-Q water. Concurrent filtration of water from up to four Niskin samplers reduced filtration time to 1 h (filtering under vacuum was avoided to reduce the possibility of cell breakage). The filtration step was intended to reduce the abundance of bactivores and larger photoautotrophs, although no measurements were made of the abundance of particles larger than bacteria. Filtrate was collected into an acid-cleaned poly-
D.A. Hansel1 et al. /Marine
carbonate carboy and then dispensed into 0.5 or 1.0 1 Pyrex’ bottles, which had been acid-cleaned (with a 5% u/u HCl solution) and precombusted at 450°C. A small headspace was left in the bottle. The incubation bottles were fitted with ground glass stoppers and sealed with Apiezon’ L high vacuum grease to prevent degassing or ingassing of the sample. Care was taken to prevent the grease from coming into contact with the water (only the top of the stopper was greased). All incubation bottles were stored in the dark, with samples collected from I 200 m depth maintained at room temperature (N 20°C; within a few degrees of the temperature at the time of sampling) while deeper samples were refrigerated at 5°C. At the start of an incubation, and at various times up to 101 days, replicated control and incubated bottles (2-3 each) were terminated by adding 100 ~1 of saturated HgCl,. The stoppers were regreased, taped to provide positive closure of the bottle and stored in the dark until analysis for TCO,. At times, separate 1.0 1 Pyrex” bottles filled with filtrate from 20 m depth were incubated in order to monitor changes in bacterial abundance and growth. These bottles were not sealed with Apiezon” grease. DOC samples were collected from the grease-free bottles and stored in precombusted glass ampoules, which were subsequently flame sealed and frozen, or in Whatman EPA precleaned vials (40 ml), with teflon-lined caps, and frozen. Time points of the incubations were chosen with regard to the requirements of the experiment. Time courses of 0, 1, 2, and 3 days were chosen to investigate rapid changes in TCO, while longer time points of 6, 10 and 76 or 101 days were chosen to investigate long-term DOC mineralization. The poisoned time-zero samples served as controls for abiotic changes in TCO, during the experiments. These samples were maintained in the same storage conditions as the live samples and were analyzed at the end of the full incubation period. Time points between 0 and 1 days were not made due to the inability of the gas extraction-coulometric procedure to clearly distinguish < 0.3 PM changes in TCO, (see below). Throughout the text the terms labile, semi-labile and refractory DOC are used. The definitions used here are: labile DOC has turnover times of hours to days; semi-labile DOC has turnover times of weeks
Chemistry 51 (1995) 201-212
to months; refractory many years. 2.2. Geochemical
203
DOC has turnover
times
of
measurements
The introduction and development of the gas extraction/coulometric analytical technique for determining TCO, (Johnson et al., 1985,1987) has greatly improved our ability to follow metabolic changes in the production and consumption of TCO,. The precision of shipboard and laboratory measurements of TCO, can be better than 1 PM (Robinson and Williams, 1991; Johnson et al., 1993; Bates et al., 1995). A SOMMA (single-operator Multiparameter Metabolic Analyzer), similar to that described by Johnson et al. (1993), was used to control the pipetting and extraction of a seawater sample. The SOMMA was interfaced with a personal computer and coupled to a CO, coulometer detector (Model 5011; UIC Coulometrics Inc.). In the method, a calibrated volume of seawater was acidified with 8% u/u phosphoric acid, converting HCO; and CO:- to undissociated CO,. Free CO, was extracted by an inert carrier gas, such as ultra high purity nitrogen or helium gas, and absorbed by a coulometer cell solution containing ethanolamine, dimethylsulfoxide and a thymolphthalein indicator. Reaction of CO, with ethanolamine forms hydroxyethylcarbamic acid which is titrated with electrolytically generated hydroxide. The coulombs of electricity used to generate hydroxide are related to the moles of CO, absorbed by the solution using the Faraday constant (Johnson et al., 1993). The measurement was calibrated with known volumes of pure CO, (Wilke et al., 1993). This method offered highly precise and rapid (12- 15 min) sample analyses. The design of the SoMMA-coulometer system was made to facilitate the measurement of metabolic processes (Johnson et al., 1993). However, considerable variability in the day-to-day and cell-to-cell response of the coulometer solutions make it difficult to resolve changes in TCO, of less than 1 PM, especially under shipboard operation (SOMMA workshop, Seattle, USA. September 1993). Efforts were therefore made to improve the resolution of the extraction-coulometric technique (Bates, 1995; Bates et al., 1995). TCO, analyses were always conducted in the same climate controlled laboratory at the
D.A. Hansell et d/Marine
204
Bermuda Biological Station for Research and analyses of mineralization samples occured only when optimal operating conditions were met. TCO, analysis was typically made on the samples within a few weeks of the last bottle being terminated. Each time course of mineralization samples (e.g. 0, 1, 2, 3, 6, and 10 days) from one depth were analyzed on the same day using the same coulometer cell solutions. This procedure was followed in order to eliminate variability in TCO, analysis on different days due to slightly variable coulometer cell conditions. The coefficient of variation for TCO, analyses were typically better than 0.02% (0.4 /IM) for within- and between-bottle replicate analyses (for both mineralization experiments and water column measurements of TCO, at the BATS site>, a factor of 6 improvement over DOC analysis. The mean difference between duplicate analyses from the same bottle was approximately 0.015% (0.35 PM) for samples analyzed at the BATS site between 1992 and 1994 (Bates, 1995; Bates et al., 1995). Concentrations of DOC were determined by high-temperature combustion, similar to the method described by Hansel1 (1993) and Hansel1 et al. (1993). The quartz combustion tube (13 mm O.D.), maintained at 700°C contained 15 g Cuprox (Leeman Labs, Inc.), 15 g Sulfix (Wake Pure Chemical Industries, Inc.) and 3 g platinum pillows (Ionics, Inc.). Carbon dioxide was detected using a LICOR Model LI-6252 NDIR analyzer. The mean of the standard deviations from 36 individual determinations in this work (3-4 injections per determination) was 2.7 PM C. Calibrations were performed daily with a 4 point standard curve using glucose (maximum concentration of 100 PM C). The system blank (normally 7-8 PM C> was determined using the vialed Millipore Milli-Q water produced at the Bermuda Biological Station for Research. This water was cross referenced to the low-carbon reference water distributed by J. Sharp (University of Delaware) as part of the community DOC measurement intercalibration in 1994.
Chemistry 51 (1995) 201-212
Porter and Feig (1980). Bacterial growth was measured by the incorporation rate of methyl-tritiated thymidine (Fuhrrnan and Azam, 1982) using a conversion factor of 1.7 X 1018 bacterial cells per mole thymidine. Reference has been made to unpublished rates of primary productivity, measured by 14C uptake, and concentrations of particulate organic carbon (POC) measured in 1993 and 1994 that were determined by the BATS program. The methods used for all of the biological determinations are given in the BATS Methods Manual (Knap et al., 1993). 2.4. Tests for incubation
Tests were performed to evaluate the experimental design for artefacts. If the sampling technique introduced labile organics there should have been a noticable stimulation of the normally slow metabolic rates of bacteria at that depth. Incubations were performed twice on water collected from 1000 m and treated as described above (prefiltration). In the first experiment, the incubation was allowed to proceed for 6 days. A second experiment was run with water from 1000 m for 76 days as a check on the results of the first, short incubation. Also evaluated was the effect of filtration on bacterial abundance, bacterial growth, and DOC concentrations. 2.5. Test of precision To evaluate the limits of precision inherent to the experimental design, twelve incubation bottles were identically filled with water collected from 20 m (Hydrostation ‘ ‘S” site, June 1993), which had been prefiltered and dispensed as described above. Six of the bottles were “killed” immediately (described above) while the remainder were allowed to incubate for 5.4 days.
3. Results 3.1. Incubation
2.3. Biological
artefacts
artefacts
measurements
Bacterial abundance was determined by epifluorescence microscopy using 4’,6-diamidino-2-phenylindole (DAPI) as a stain according to the method of
Several tests were performed to evaluate the experimental design for artefacts, such as unintended stimulation or inhibition of bacterial respiration in the incubation bottles. Stimulation could occur due
D.A. Hansel1 et al. /Marine 201
I
04 . 0
I
2
.
I
I
I
I
4
6
8
10
Time
.
1
12
.
1
14
.
I
1
16
(days)
Fig. 1. Bacterial abundance (X 10’ 1.’ ) in an incubation container as a function of time in water collected from 20 m (May 1993) and filtered (0.8 pm pore size). One filter was counted per sample. The error bars are the standard deviation of 15 squares (each 2000 prn2) counted per filter.
to the introduction of labile organic carbon to the system (such as unintentional contamination with the Apiezon’ grease used to seal the bottles) or from the reduction of grazing pressure on the microbes. Inhibition of microbial metabolism could occur from physical damage to the organisms incurred with filtration or from the introduction of toxic compounds. Filtration of water collected from 20, 80 and 200 m (April 1994) did not result in a measurable increase in the DOC concentration (concentration stayed within f2 PM). Bacterial numbers were reduced by 11% with filtration (November 1993). In a separate test (May 1993; no concurrent TCO, production measurements were made), bacterial numbers doubled in the filtered samples over the first two days of the incubation (Fig. l), and decreased between days 6 and 16. Kirchman et al. (1991) observed similar bacterial abundance patterns, with a reduction of bacterial numbers after 2-3 days as the number of bactivore flagellates increased. In April 1994, initial bacterial growth rates in 20 m water was only 27% of growth in unfiltered seawater, but growth rates returned to near ambient levels in the course of a day (Fig. 2). Growth of bactivores in incubation chambers has been shown to occur after the first few days of incubation (Kirchman et al., 1991) but bactivore abundance was not monitored in this work. Because
Chemistry 51 (1995) 201-2 I2
205
of the anticipated lag in their appearance, the first few time points reported here should be good estimates of bacterial respiration. For the longer incubations (lo- 100 days), the presence of bactivores may improve the estimates of DOC utilization by bacteria. If marine bacteria have an assimilation efficiency of 10% (see below), then that percentage of DOC processed would be retained as bacterial biomass. Introduction of higher trophic levels such as bactivores should, in effect, increase mineralization of the DOC pool because they will respire a large fraction of the bacterial carbon ingested. TCO, production then becomes a strong indicator of total DOC utilization by bacteria over the long term. Finally, the assumption is made in this work that the respiration measured is not due to direct DOC utilization by microheterotrophs. If such a process is important in nature then the short-term rates reported may underestimate total DOC utilization because the organisms larger than 0.8 pm in size were intentionally removed. 3.2. Test of precision
of the protocol
This test was designed to quantify the variation in TCO, signal due to the filling of the bottles plus 1.0 7
t
0.8I d 5 % $ 3 s z ii
0.6-
0.4-
t 0.2-.
0.04 0
I
1
4
8
.
I
12
Time
I
16
I
20
I
24
28
(hours)
Fig. 2. Bacterial growth, measured as the incorporation rate of tritiated thymidine (pmole Tdr L-’ h”), in 0.8 pm filtered water (a). The water was taken from 20 m (April 1994). For comparison, the incorporation rate of unfiltered water at the start of the experiment is shown (0). Error bars are the standard deviations of the measurements (a = 3). The number of bacteria did not change significantly (P > 0.05; Student’s t-test; Sokal and Rohlf, 1969) over the time course of 1 day.
206
D.A. Hansel1 et d/Marine
Chemistry 51 (1992) 201-212
analytical limitations, as well as the variability imparted by the biological processes occurring over the period of incubation. The variability in the concentrations of TCO, in the six initial and final bottles represent the summed uncertainty associated with the protocol (Table 1). Initial bottles had a mean TCO, concentration of 2060.80 PM, with a standard deviation of 0.62 PM (CV = 0.030%), while the final bottles had a higher mean concentration of TCO, (2063.01 PM) and a larger standard deviation (1.13 ,uM; CV = 0.055%). The precision of the TCO, concentration determined for the initial bottles reflects the variability associated with filling the bottles and analytical procedures. The variability from the final bottles reflects the additional natural biological variability. The mean rate of TCO, production in these bottles was 0.41 PM dd ‘. 3.3. Short-term
0
2
3
4
Time (days) Fig. 3. Time courses of the production of TCO, (as proxy for the mineralization of DOC) during short-duration incubations. Error bars are the standard deviations of the measurements.
incubations
One-day incubations were performed to determine the initial rates of TCO, production resulting from the microbial mineralization of DOC. Three incubations were performed in the period from September 1993 until April 1994, each with water collected from 20 rn at the BATS or Hydrostation “S” sites. In the three experiments, the rate of TCO, production after 24 h was essentially identical at 0.440.45 FM d-’ (Fig. 3; Table 2). During the September time course the rate of TCO, production remained linear for the first three days of the incubation, while in all other cases the rates slowed beTable 1 Concentrations
1
tween days 1 and 3 (i.e. 0.35-0.45 PM d-’ at 2 days and 0.28-0.44 PM d-’ at 3 days). Incubations of intermediate duration (9 days) were performed with water collected from 3 depths in the upper 200 m of the water column during May 1993. The fastest initial rate of TCO, accumulation (3 day time point) was in the 20 m water, with rates at 70 and 200 m identical to one another (Fig. 4). The rates slowed after day 6, except at 70 m where TCO, production continued at a rate similar to the earlier time points. After 9 days, the amount of TCO, produced was similar at 20 and 70 m (1.36 and 1.57
of TCO? ( WM C) in the six initial and six final incubation
Initial
bottles. filled in June 1993
Final
Bottle
TCO,
Bottle
TCO,
1
2061.57
3 5 7 9 11
2061.35 2060.65 2060.23 2060.06 2060.83
(0.25) (0.00) (*I (0.31) (0.22) (0.12)
2 4 6 8 10 12
2064.19 2062.91 2064.06 2063.39 2061.56 2061.89
(0.43) (0.11) (0.47)
Grand Mean
2060.80
(0.62)
Grand Mean
2063.01
(1.13)
(0.48)
Final bottles were poisoned at 5.4 days. The concentration reported from each bottle is the mean of duplicate TCO, determinations (standard deviations in parentheses). Grand means and standard deviations were calculated from the eleven individual TCO, analyses performed in each treatment. * = Single determination from bottle.
D.A. Hansel1 et al./Marine
201
Chemistry 51 (1995) 201-212
Table 2 Rates of TCO, production ( PM C dd’ ; assumed equivalent to microbial respiration) and primary productivity (PP; PM C d- ’ ). All rates are from 20 m at the BATS site. Bacterial growth f PM C d- I >wascalculated using the conversion factor of 1.7X IO’* bacterial cehs per mole thymidine (Carlson et al., 199.5). A carbon conversion factor of 9.2 fg C (bacterial celI)- ’ was used (K. Gundersen, unpubl. data) Date September 1993 November 1993 April 1994
TCO,
PP
TCO, X PP-’
Bacterial growth
Efficiency
0.45 0.44 0.45
0.42 0.40 0.20
1.07 1.10 2.25
0.044 0.033 0.018
8.9 7.0 3.8
PM, respectively), while the amount produced at 200 m was 0.97 PM. Bacterial growth efficiencies were calculated from the bacteria1 growth rate, measured by thymidine uptake, and from the production of CO, (Table 2). Efficiencies ranged from 3.8 to 8.9%, with a mean of 6.2%. This range is in the low end of published efficiencies compiled by Jahnke and Craven (1995) and Carlson and Ducklow (1995), but similar to that estimated by Carlson and Ducklow (1995) for the BATS site. 3.4. Long-term
incubations
To determine the amount of DOC available to the microbes over periods of a few months, incubations were carried out for 76 or 101 days on water collected from 20, 80, 200 and 1000 m. The rate of DOC mineralization in samples from the mixed layer
was most rapid in the first 10 days of the incubation, and then slowed (Fig. 5). The size of the biologically available DOC pool (time scale of weeks to months) was variable, ranging from 3.3 to 6.6 FM C (Table 3) in the upper 200 m of the water column. DOC mineralization was not detectable in deep waters (1000 m) after 76 days. This last result indicates that inadvertent contamination of the samples prior to incubation can be avoided when caution is exercised. No clear correlation between the amount of semilabile DOC and depth or season was evident in the data presented in Table 3. As would be expected, the concentration of biologically available DOC at 1000 m was non-detectable. The DOC concentration at 1000 m, then, indicates the base concentration above which labile and semi-labile DOC would accumulate in the surface layer. The size of the labile plus semi-labile pool is ultimately defined by the difference in concentrations between the deep and surface layers.
I
-0.5 4
0
(%)
2
4
6
8
10
12 -I
Time (days) Fig. 4. Time courses of the production of TCO, (as proxy for the mineralization of DOC) during incubations of intermediate duration. Error bars are the standard deviations of the measurements.
20
40
60
80
100
120
Time (days) Fig. 5. Time courses of the production of TCO, (as proxy for the mineralization of DOC) during long-duration incubations.
D.A. Hansel1 et al. /Marine
208
Table 3 TCO, accumulation (PM) following incubations of water collected at the BATS and Hydrostation “S” (January 1994 only) sites, and the ambient DOC concentrations ( PM) Date
Depth(m)
TCO,
DOC
Duration (d)
November 1993 November 1993 November 1993 January 1994 January 1994
20 80 1000 20 200
4.25 3.31 nd 6.64 4.41
na na 46.2 b 69.6 = 72.6 a
76 76 76 101 101
Prefiltration was as described in the text. ’ Concentrations after 10 days of incubation. Initial samples had been contaminated during ampoulation. Values will underestimate the true concentrations by < 4 WM. b The mean of the concentrations of DOC at 1000 m (n = 5; standard deviation = 2.2 PM) determined from February-June, 1994. na = not available; nd = non-detectable.
4. Discussion The goals of the experiments were to determine the “instantaneous” rate of DOC mineralization by bacteria and to determine the quantity of biologically available DOC, on the time scale of weeks to months. To accurately determine the “instantaneous” rate, several requirements must be met: (1) the respiration measured must primarily be that of the organisms utilizing DOC as substrate; (2) there must be no stimulatory or inhibitory effects due to sample manipulation and; (3) the incubations have to be of short enough duration that the rates are representative of natural levels. Bacteria are the primary consumers of DOC in marine systems and they are the primary group of organisms that we expect to pass through the filters used. Keil and Kirchman (1991) found in the subarctic Pacific that 0.8 pm filtration passed 97% of the bacterial assemblage, but only 0.5% of the chlorophyll. This result suggests that most of the uptake of dissolved primary amines they measured was due to bacteria. In our work, 89% of the bacteria at 20 m passed through the filter. In the Sargasso Sea, bacterial biomass may make up a very large portion of the euphotic zone microbial carbon (Fuhrman et al., 1989), so passage of a fraction of the autotrophic assemblage should not have greatly impacted the rates we measured. We were unable to resolve an increase in the
Chemistry 51 (1995) 201-212
concentrations of DOC due to filtration, but studies using more sensitive techniques have shown that some DOC should be expected to be released from particles during filtration (Goldman and Dennett, 1985). Keil and Kirchman (1991), performing similar gravity filtrations using 0.8 pm polycarbonate filters, found that dissolved free amino acid concentrations increased by an average of 11%. Hilmer and Bate (1989) reported a total photosynthate release of only 3%, this value apparently including naturally released, 14C-labeled DOC. Goldman and Dennett (1985) and Hilmer and Bate (1989) focused on recently fixed carbon, and not on the total mass of particulate organic carbon (POC) released as DOC during filtration. The magnitude of the total release is likely a function of the POC load, which in the Sargasso Sea is low (Michaels et al., 1994). During the period when water was collected for the experiments reported here, the ambient concentrations of POC at 20 m had a narrow range of 1.6-2.0 PM C (or < 3% of the DOC concentration). If it is assumed that 5% of the POC pool (exclusive of bacteria) can be passed through a filter and made available as DOC, then the increase to the ambient DOC pool was no more than 0.08-o. 1 PM C. This level of increase must be viewed as a “trace” addition to the biologically available DOC pool, which is shown below to be an order of magnitude or more larger in the Sargasso Sea. If a stimulation of the bacterial population occured at this level, it was below the level of sensitivity available for resolving its contribution. Rationalization on why the sampling procedure was not thought to stimulate bacterial metabolism has been presented. It appears, though, that for several hours immediately following filtration, bacterial growth rates may be reduced (Fig. 2). Growth rates returned to levels near the initial, unfiltered value within 25 h of the start of the incubation. The reductions in growth rate were contrary to the results of Coffin et al. (1993) who found the highest rates of microbial respiration during the hours immediately following filtration. Studies on bacterial respiration and growth have suggested that bacteria are rapidly respiring a very small pool of highly labile, recently produced DOC (Fuhrman, 1987). Coffin et al. (1993), for example, working in near-shore waters, showed a reduction in the rate of oxygen consumption 5- 12 h
D.A. Hansel1 et al./Marine
after 1.0 pm filtration, apparently due to exhaustion of the most labile fraction of DOC. Application of the method reported here in oligotrophic waters cannot resolve changes over short time scales, leaving open the possibility that the rates based on 24 h incubations are underestimates. In order to better assess the validity of the rates of DOC mineralization we report, the rates can be compared to the few other reports of microbial respiration or DOC consumption in the Sargasso Sea. Kepkay et al. (1990) measured respiration in water collected from 10 m and prefiltered through a 2 pm filter. Mean oxygen consumption was 0.37 /_LM 0, dd’ (or 0.27 PM C d-‘, using their molar respiratory quotient of 1.35) while primary productivity was 0.38 PM dd’. The rate of DOC mineralization they reported was 40% lower than our value at 20 m. The mean of the whole water (20-25 m) respiration rates reported by Williams and Jenkinson (1982) was 1.24 PM C d-‘, or 2.7 times the rate we measured. Bumey (1986) determined the uptake of dissolved carbohydrate at 30 m to be 0.12 PM C h- ’ while Carlson and Ducklow (1995) reported a range of 0.04-0.10 PM C hh’ of DOC utilization at 10 m. Much of the labile plus semi-labile DOC in the surface layer of the ocean is likely composed of carbohydrates (Benner et al., 1992) so Bumey’s results should be directly comparable to the bulk measurements of DOC utilization. An unexpected result of our work was that the rates of DOC mineralization were high relative to concurrent rates of primary productivity (Table 2). If the mineralization of DOC is indeed tightly coupled to its production, then it follows that the rate of microbial respiration should not ordinarily be equal to or greater than the gross rate of primary productivity. This necessity stems from other loss terms expected, such as the grazing of the daily primary production in the Sargasso Sea that must take place. Roman et al. (1993) reported grazing rates in August of 0.6 mg C mP2 h- ’ for mesozooplankton. No published data have been found for microzooplankton grazing on phytoplankton in the Sargasso Sea but their impact is likely significant. So, in order for near-balance between the measured rates of DOC consumption and measured primary productivity to exist then: (1) there must exist significant pools of semi-labile DOC, the utilization of which pushes the
Chemistry 51 (1995) 201-212
209
system into long periods of net heterotrophy, or; (2) the measured primary productivity underestimated gross primary productivity at the time of the experiments. 4.1. Sources of semi-labile
DOC
Two source terms for the semi-labile DOC pool, independent of recently produced photosynthate, can be identified. These are the photo-oxidation of refractory DOC, of a deep-water source, in the upper few meters of the water column and the utilization of DOC produced during the previous spring bloom or some other period of high DOC production, which is semi-refractory to microbial utilization and slowly mineralized. Recent work has demonstrated that the photo-oxidation of biologically refractory DOC, mixed to the surface during winter overturn, may be an important sink for deep-water DOC (Kieber et al., 1989; Mopper et al., 1991). This deep-water pool of DOC has been characterized as refractory to microbial consumption (Barber, 1968; Table 3). Mopper et al. (1991), however, measured the production of low molecular weight carbonyl compounds in the surface waters of the Sargasso Sea (4.1 nM C h-‘, exclusive of CO production they measured). They implicated natural levels of UV light as the causative agent and suggested deep-water DOC as the source. If this rate is doubled to account for other low molecular weight organics not measured (e.g. organic acids), and 8 h dd’ of adequate light is assumed, then the daily rate of production of low molecular weight, labile DOC may be 0.07 PM d.- ‘. The rate of production should be equal to its utilization by bacteria. The extent to which this process occurs at 20 m, the shallowest depth examined in our work, is unknown. The second source of semi-labile DOC is that produced and accumulated during the Sargasso Sea spring bloom or during other periods of high DOC production. The region represented by the BATS site is characterized by a strong seasonality in mixed layer depths, a winter recharge of surface nutrients and a vernal phytoplankton bloom (Lohrenz et al., 1992; Michaels et al., 1994). The 1993 spring bloom, the bloom preceding our experiments, reached levels of primary production as high as 880 mg C mm 2 d-‘. Unlike the clear spring bloom in 1993, there was no major bloom in 1994, apparently due to the
210
D.A. Hansel1 et al. / Marine Chemistry 51 (1995) 201-212
absence of wintertime convective overturn of the water column. Rates of primary production remained below 400 mg C mm2 dd’ throughout the 1994 winter/spring period. The “short-term” rates that we report in Fig. 3 were determined in fall, following the 1993 bloom, and the winter and spring of 1994. Carlson et al. (1994), reporting seasonal variability in the concentrations of DOC at the BATS site, found an accumulation of DOC above background concentrations (0.6-0.8 mol C mP2 above late summer values) during two consecutive spring blooms. The concentrations of DOC decreased following the bloom periods, in both the upper water column (O-100 m) and in the layer immediately below (loo-250 m). Decreases in DOC concentrations in both layers may approximate net rates of mineralization (changes in concentration due to advection could not be resolved). The timing of DOC accumulation in the deep layer ties its production to the spring bloom. Strong seasonal stratification following the spring bloom effectively seals the deeper layer, largely eliminating additional inputs of DOC from the euphotic zone by mixing. The consumption of DOC in the lower layer after stratification should represent the mineralization of semi-refractory DOC produced during the spring bloom. The mean rate of decrease in the DOC concentration (loo-250 m) from March to April 1993 in the Carlson et al. (1994) data was 0.17 PM dd ’ . The rate of DOC disappearance decreased thereafter. This finding is in reasonable agreement with the DOC mineralization rate of 0.07 PM dd’ measured in May, 1993 at 200 m (Fig. 4). We suggest that DOC mineralization occuring below the seasonal thermocline may largely be supported by substrate produced during the previous spring bloom. If due to the spring bloom, then the rate of mineralization below the seasonal thermocline should be indicative of the microbial consumption of the same fraction of DOC (residual from the spring bloom) in the surface mixed layer. For budgeting purposes at 20 m, the rate of DOC mineralization in the deep layer (at 200 m) is taken to represent the rate of mineralization of the same fraction of semi-refractory DOC residing in the euphotic zone. That rate is 0.07 PM d-l, or 15% of the total mineralization rate (0.45 FM C d-l) at 20 m. In
balance, 85% of the daily rate of DOC mineralization rate must be supported by recent primary production (the contribution of photo-oxidation at 20 m is assumed to be negligible). This result requires that 0.38 PM C d-’ (or 90%) of the daily primary production (0.42 PM dd ’ ; Table 2) measured in September 1993, for example, be consumed by bacteria. This value is somewhat greater than that of Biddanda et al. (1994), who found that 69% of the daily primary production in the slope waters of the northern Gulf of Mexico was processed by the bacterioplankton. 4.2. Underestimation
of gross primary productiuity
We have no evidence that there is a significant difference between net primary production measured in the BATS program (here defined as the 14C activity in filtered particles, corrected with dark incubations, after 24 h, single-time point incubations) and gross primary production (here defined as summed fixation of carbon into the particulate and dissolved organic pools). There remains, though, a lingering concern that in many areas the 14C method for primary production underestimates the true production by large amounts (Banse, 1994). If gross primary production is indeed underestimated then much of the underestimate might be found in the dissolved organic pool. Lancelot (1979) found 14C to accumulate into the DOC pool at rates approximately l/3 of the accumulation in particles, while Laws et al. (1984) found accumulation of labelled DOC ranged from non-detectable to 18% of the particulate pool after 24 h incubations. The transfer of label from the particulate to the dissolved pool these studies report can be large, but they are not on the order of equivalence between net production and DOC consumption as required by the results of this work. The few rates of DOC consumption reported for the Sargasso Sea (listed above) are, like those of this work, high relative to anticipated concurrent levels of primary productivity (rates at the BATS site range from 2 to 20 mg C me3 dd ‘; BATS Data Reports). Note, though, that the 14C method of determining DOC production traces the transfer of recent photosynthate to solution, while measurements of microbial utilization of DOC encompass all DOC source terms including, but not limited to, recent photosyn-
D.A. Hansel1 et al./Marine
thate. So the discrepancy that exists (i.e. the ratio of 14C estimated DOC production to primary productivity, as reported in the literature, is generally smaller than the expected ratios of DOC mineralization to primary productivity in the Sargasso Sea) may be due to processes that contribute to the production of labile or semi-labile DOC but that cannot be classified as loss of recent photosynthate. Some processes have been hypothesized (sloppy feeding, loss of DOC from egested fecal pellets, dissolution of sinking particles, viral disruption of bacteria) but none have been adequately quantified. These secondary processes of DOC production clearly need to be evaluated for their role in supporting bacterial growth.
Acknowledgements We thank Drs. A.F. Michaels and A.H. Knap for access to the BATS data from 1993 and 1994. Support for DAH came from NSF grants OCE9311012 and OPP-9317200, while support for NRB and KG came from NSF grant OCE-9301950. J.1 Hedges, C.A. Carlson, E.T. Peltzer, R.G. Keil and one anonymous reviewer are thanked for their helpful and constructive comments. The US Department of Energy CO, Science Team is acknowledged for the loan of a SoMMA-Coulometer TCO, Analyzer to BBSR. This is Contribution No. 1404 from the Bermuda Biological Station for Research, Inc.
References Amon, R.W. and Bemier, R., 1994. Rapid cycling of high-molecular weight dissolved organic matter in the ocean. Nature, 369: 549-552. Bacastow, R. and Maier-Reimer, E., 1991. Dissolved organic carbon in modeling oceanic new production. Global Biogeochem. Cycles, 5: 71-85. Banse, K., 1994. Commentary: An unpleasant note about the carbon-14 method of estimating plankton photosynthesis. US JGOFS Newslett., 6: 3-15. Barber, R.T., 1968. Dissolved organic carbon from deep water resists microbial oxidation. Nature, 220: 274-275. Bates, N.R., 1995. Investigation of the Physical and Biological Controls of the Oceanic CO, System in the Sargasso Sea. Ph.D. Dissertation. Univ. Southampton, Southampton, 277 pp. Bates, N.R., Michaels, A.F. and Knap, A.H., 1995. Seasonal and interannual variability of the oceanic carbon dioxide system at
Chemistry 51 (1995~ 201-212
211
the U.S. JGOFS Bermuda Atlantic Time-Series Site. Deep-Sea Res. 11, in press. Benner, R., Pakuiski, J.D., McCarthy, M., Hedges, J-1. and Hatcher, P.G., 1992. Bulk chemical characteristics of dissolved organic matter in the ocean. Science, 255: 1561- 1564. Biddanda, B., Opsahl, S. and BeMer, R., 1994. Plankton respiration and carbon flux through bactetioplankton on the Louisiana shelf. Limnol. Oceanogr., 39: 1259-1275. Bumey, C.M.. 1986. Bacterial utilization of total in situ dissolved carbohydrate in offshore waters. Limnol. Oceanogr., 31: 427431. Carlson, C.A. and Ducklow, H.W., 1995. Bacterial growth and dissolved organic carbon utilization in the oligotrophic Sargasso Sea. Aquat. Microb. Ecol., submitted. Carlson, CA., Ducklow, H.W. and A.F. Michaels, 1994. Annual flux of dissolved organic carbon from the euphotic zone in the northwestern Sargasso Sea. Nature, 371: 405-408. Carlson, C.A., Ducklow, H.W., Sleeter, T.D., 1995. Stocks and dynamics of bactetioplankton in the northwestern Sargasso Sea. Deep-Sea Res., in press. Coffin, R.B., Connolly, J.P. and Harris, P.S., 1993. Availability of dissolved organic carbon to bactetioplankton examined by oxygen utilization. Mar. Ecol. hog. Ser., 101: 9-22. Ducklow, H.W. and Carlson, C.A., 1992. Oceanic bacterial production. Adv. Microb. Ecol., 12: 113-181. Eppley, R.W., Honigan, S.G., Fuhrman, J.A., Brooks. E.R., Price, C.C. and Sellner, K., 1981. Origins of dissolved organic matter in Southern California coastal waters: experiments on the role of zooplankton. Mar. Ecol. Prog. Ser., 6: 149-159. Fuhrman, J.A. and Azam, F., 1982. Thymidine incorporation as a measure of heterotrophic bacterial production in marine surface waters: evaluation and field results. Mar. Biol., 66: 109-120. Fuhrman, J.A., 1987. Close coupling between release and uptake of dissolved free amino acids in seawater studied by an isotope dilution technique. Mar. Ecol. Prog. Ser., 37: 45-52. Fuhrman, J.A., Sleeter, T.D., Carlson, C.A. and Proctor, L.M., 1989. Dominance of bacterial biomass in the Sargasso Sea and its ecological implications. Mar. Ecol. Prog. Ser., 57: 207-217. Goldman, J.C. and Dennett, M.R., 1985. Susceptibility of some marine phytoplankton species to cell breakage during filtration and post-filtration rinsing. J. Exp. Mar. Biol. Ecol., 86: 47-58. Hansell, D.A., 1993. Results and observations from the measurement of DOC and DON using high temperature catalytic combustion techniques. Mar. Chem., 41: 195-202. Hansell, D.A., Williams, P.M. and Ward, B.B., 1993. Measurements of DOC and DON in the Southern California Bight using oxidation by high temperature combustion. Deep-Sea Res., 40: 219-234. Hedges, J.I., 1992. Global biogeochemical cycles: Progress and problems. Mar. Chem., 39: 67-93. Hilmer, T. and Bate, G.C., 1989. Filter types, filtration and post-filtration treatment in phytoplankton production studies. J. Plankt. Res., 11: 49-63. Holmes, R., Williams, P.M. and Eppley, R.W., 1967. Red water in La Jolla Bay. Limnol. Oceanogr., 12: 5033512. Ignatiades, L. and Fogg, G.E., 1973. Studies on the factors
212
D.A. Hansel1 et al./Marine
affecting the release of organic matter by Skeleronema costarum (Greville) Cleve in culture. J. Mar. Biol. Assoc. UK, 53: 937-956. Jahnke, R.A. and Craven, D.B., 1995. Quantifying the role of heterotrophic bacteria in the carbon cycle: A need for respiration rate measurements. Limnol. Oceanogr., 40: 436-441. Johnson, K.M., King, A.E. and Sieburth, J.McN., 1985. Coulometric TCO, analyses for marine studies: an introduction. Mar. Chem., 16: 61-82. Johnson, K.M., Sieburth, J.M., Williams, P.J. 1eB. and BrandStrom, L., 1987. Coulometric total carbon dioxide analysis for marine studies: automation and calibration. Mar. Chem., 21: 117-133. Johnson, K.M., Wills, K.D., Butler, D., Johnson, W.K. and Wong, C.S., 1993. Coulometric total carbon dioxide analysis for marine studies: maximizing the performance of an automated gas extraction system and coulometric detector. Mar. Chem., 44: 167-188. Jumars, P.A., Penry, D.L., Baross, J.A., Perry, M.J. and Frost, B.W., 1989. Closing the microbial loop: Dissolved carbon pathway to heterotrophic bacteria from incomplete ingestion, digestion and absorption in animals. Deep-Sea Res., 36: 483495. Knap, A.H., Michaels, A.F., Dow, R.L., Johnson, R.J., Gundersen, K., Sorenson, J.C., Close, A.R., Howse, F.A., Hammer, M., Bates, N.R., Doyle, A. and Waterhouse, T., 1993. BATS Methods Manual. US JGOFS Planning Off., Woods Hole, MA. Keil, R.G. and Kirchman, D.L., 1991. Contribution of dissolved free amino acids and ammonium to the nitrogen requirements of heterotrophic bacterioplankton. Mar. Ecol. Prog. Ser., 73: l-10. Kepkay, P.E., Harrison, W.G. and Irwin, B., 1990. Surface coagulation, microbial respiration and primary production in the Sargasso Sea. Deep-Sea Res., 37: 145-155. Keys, A., Christensen, E.H. and Krogh, A., 1935. The organic metabolism of sea-water with special reference to the ultimate food cycle in the sea. J. Mar. Biol. Assoc. UK, 20: 181-196. Kieber, D.J., McDaniel, J. and Mopper, K, 1989. Photochemical source of biological substrates in sea water: implications for carbon cycling. Nature, 341: 637-639. Kirchman, D.L., Suzuki, Y., Garside, C. and Ducklow, H.W., 1991. High turnover rates of dissolved organic carbon during a spring phytoplankton bloom. Nature, 352: 612-614. Lancelot, C., 1979. Gross excretion rate of natural marine phytoplankton and heterotrophic uptake of excreted products in the Southern North Sea, as determined by short-term kinetics. Mar. Ecol. Prog. Ser., 1: 179-186. Laws, E.A., Redalje, D.G., Haas, L.W., Bienfang, P.K., Eppley, R.W., Harrison, W.G., Karl, D.M. and Marra, J., 1984. High phytoplankton growth and production rates in oligotrophic Hawaiian coastal waters. Limnol. Oceanogr., 29: 943-948. Lohrenz, S.E., Knauer, G.A., Asper, V.L., Tuel, M., Michaels, A.F. and Knap, A.H., 1992. Seasonal variability in primary production and particle flux in the northwestern Sargasso Sea:
Chemistry S1 (1995) 201-212 U.S. JGOFS Bermuda Atlantic Time-series Study. Deep-Sea Res., 39: 1373-1391. Mackenzie, F.T., 1981. Global carbon cycle: Some minor sinks for CO,. In: G.E. Likens, F.T. Mackenzie, J.E. Richey, J.R. Sedell and K.K. Turekian (Editors), Flux of Organic Carbon by Rivers to the Oceans. US Dep. Energy, Washington, DC, pp. 360-384. Michaels, A.F., Knap, A.H., Dow, R.L., Gundersen, K., Johnson, R.J., Sorensen, J., Close, A., Knauer, G.A., Lohrenz, SE., Asper, V.A., Tuel, M. and Bidigare, R., 1994. Seasonal patterns of biogeochemistry at the U.S. JGOFS Bermuda Atlantic Time-series Study site. Deep-Sea Res., 41: 1013-1038. Mopper, K., Zhou, X., Kieber, R.J., Kieber, D.J., Sikorski, R.J. and Jones, R.D, 1991. Photochemical ‘degradation of dissolved organic carbon and its impact on the oceanic carbon cycle. Nature, 353: 60-62. Nagata, T. and Kirchman, D.L., 1992. Release of macromolecular organic complexes by heterotrophic marine flagellates. Mar. Ecol. Prog. Ser., 83: 233-240. Najjar, R.G., Sarmiento, J.L. and Toggweiler, J.R., 1992. Downward transport and fate of organic matter in the ocean: simulations with a general circulation model. Global Biogeochem. Cycles, 6: 45-76. Ogura, N., 1972. Rate and extent of decomposition of dissolved organic matter in surface seawater. Mar. Biol., 13: 89-93. Porter, K.G. and Feig, Y.S., 1980. The use of DAPI for identifying and counting aquatic microflora. Limnol. Oceanogr., 25: 943-948. Robinson, C. and Williams, P.J. leB., 1991. Development and assessment of an analytical system for the accurate and continual measurement of total dissolved inorganic carbon. Mar. Chem., 34: 157-175. Roman, M.R., Dam, H.G., Gauzens, A.L. and Napp, J.M., 1993. Zooplankton biomass and grazing at the JGOFS Sargasso Sea time series station. Deep-Sea Res., 40: 883-901. Sokal, R.R. and Rohlf, F.J., 1969. Biometry. Freeman, New York, NY, 859 pp. Taylor, G.T., Iturriaga, R. and Sullivan, C.W., 1985. Interactions of bactiverous grazers and heterotrophic bacteria with dissolved organic matter. Mar. Ecol. Prog. Ser., 23: 129-141. Toggweiler, J.R., 1989. Is the downward flux of dissolved organic matter (DOM) important in carbon transport? In: W.H. Berger, V.S. Smetacek and G. Wefer (Editors), Productivity of the Ocean: Present and Past. Wiley, Chichester, pp. 65-83. Wakesman, S.A. and Carey, C.L., 1935. Decomposition of organic matter in sea water by bacteria. II. Influence of addition of organic substances upon bacterial activities. J. Bacterial., 29: 545-561. Wilke, R., Wallace, D.W.R. and Johnson, K.M., 1993. A waterbased, gravimetric method for the determination of gas sampling loop volume. Anal. Chem., 65: 2403-2406. Williams, P.J. 1eB. and Jenkinson, N.W., 1982. A transportable microprocessor-controlled precise Winkler titration suitable for field station and shipboard use. Limnol. Oceanogr., 27: 576-584.
