Bacterial utilization of different size classes of dissolved organic matter

Transcription

1 Limnol. Oceanogr., 41(l), 1996, Q 1996, by the American Society of Limnology and Oceanography, Inc. Bacterial utilization of different size classes of dissolved organic matter Rainer M. W. Arnon and Ronald Benner Marine Science Institute, POB 1267, University of Texas at Austin, Port Aransas Abstract Bacterial utilization of high-molecular-weight (HMW; > 1 kda) and low-molecular-weight (LMW; < 1 kda) dissolved organic C (DOC) was investigated in freshwater and marine systems by measuring dissolved oxygen consumption, bacterial abundance, and bacterial production in size-fractionated samples. Tangentialflow ultrafiltration was used to separate HMW and LMW DOC. More than 80% of the DOC in Amazon River samples was recovered in the HMW fraction, whereas most marine DOC (up to 70%) was of LMW. Bacterial growth efficiencies were consistently higher in the LMW fractions (16-66%) than in the HMW fractions (8-39%), indicating compositional differences in the two size fractions. In all experiments, measured rates of bacterial growth and respiration in HMW incubations were higher than those in LMW incubations. Carbon-normalized bacterial DOC utilization rates were fold greater in the HMW fractions than in the LMW fractions, and a greater proportion ( %) of HMW DOC was utilized per day than LMW DOC ( %). All bacterial growth and respiration measurements indicated that HMW DOC was utilized to a greater extent than LMW DOC in all environments investigated. The traditional model of DOM degradation, stating that LMW compounds are most bioreactive, does not appear to apply to the bulk of natural DOM. Rather, the data and results from independent studies suggest a new conceptual model whereby the bioreactivity of organic matter decreases along a continuum of size (from large to small) and diagenetic state (from fresh to old). This size-reactivity continuum model suggests that the bulk of HMW DOM is more bioreactive and less diagenetically altered than the bulk of LMW DOM. Dissolved organic carbon (DOC) in aquatic environments represents one of the largest active organic C reservoirs in the biosphere. The amount of DOC in aquatic systems is about equal to the amount of CO,-carbon in our atmosphere (Farrington 1992). It is widely accepted that dissolved organic matter (DOM) represents a dynamic component in the interaction between geosphere, hydrosphere, and biosphere and as such has the potential to influence the global carbon cycle and climatic change (Farrington 1992). Heterotrophic bacteria are considered major consumers and remineralizers of DOM in the ocean Acknowledgments We thank the scientists, captain, and crew on the RV Longhorn and RV Amanai for assistance in collecting samples and conducting experiments. We acknowledge the hospitality of the scientists at the Institut0 National de Pesquisas da Amazonia (INPA) in Manaus, Brazil, with special thanks to E. Sargentini and A. Menegario and the other staff members of the plantchemistry laboratory for their cooperation and lab space. We especially thank T. Pimental, P. Pinheiro, B. Forsberg, C. E. Gianini, and C. R. Padovani for logistical support during the research stay in the Amazon. We also thank Dean Pakulski for help with the experiments and valuable comments throughout this study, and John Hedges for valuable discussion of the sizereactivity continuum model. We thank Curtis Suttle and two anonymous reviewers for constructive suggestions which improved the manuscript. This research was supported by grants from the National Science Foundation (DEB 9 l-20038; OCE ) and the Coastal Ocean Program Office of the NOAA (NA 90AA-D- SG689). Contribution 957 from the University of Texas Marine Science Institute and 77 from the CAMREX project. (Pomeroy 1974) and represent a very dynamic compartment in global biogeochemical cycles. The interactions between DOM and bacteria play a central role in the aquatic carbon cycle; thus, the factors regulating DOM production and consumption profoundly influence carbon fluxes. The availability of DOM to heterotrophic bacteria likely depends on its biochemical composition and molecular size, inorganic nutrient concentrations, and other environmental factors, such as temperature. Molecular weight or size is often mentioned as an important factor influencing the microbial utilization of DOM. Saunders (1976) formulated a general model that simple organic molecules decompose most quickly, with- in hours. Phytoplankton derived high-molecular-weight (HMW) organic compounds decompose within days to weeks, while other HMW compounds decompose on the order of months or more. Although this conceptual model is still widely accepted, recent studies indicate that some HMW compounds are rapidly utilized by bacteria (e.g. Meyer et al. 1987; Tranvik 1990; Arnosti et al. 1994). Combining information on the average 14C-age of DOC in the ocean (Williams and Druffel 1987; Bauer et al. 1992) with observed molecular-weight distributions of DOC in aquatic environments (Carlson 1985; Benner et al ; Ogawa and Ogura 1992) raises an interesting question. If -80% of the DOC in the deep ocean is of low molecular weight (LMW; < 1 kda) with an average age of 4,000-6,000 years, how reactive is the bulk of this LMW material? Is the traditional view that LMW organic compounds are the most labile components of DOM really justified for the bulk of LMW DOM in different aquatic environments? In a recent study, Amon and Benner (1994) found that HMW DOC sampled during a diatom bloom was utilized faster and to a greater extent than the cor- 41

2 42 Amon and Benner Table 1. Sampling site locations and characteristics of the water collected as well as the initial DOC concentrations (PM C) in the various incubations. LMW- < 1 kda; HMW- > 1 kda (and in the Rio Negro and Rio Solimces experiments > 1 kda and < 30 kda); VHMW- > 30 kda. Location Rio Negro Rio SolimGes Laguna Madre Aransas Pass,Gulf of Mexico Gulf of Mexico Gulf of Mexico Description High-water season High-water season Seagrass-dominated estuary Ship channel, incoming tide Cyanobacteria Atchafalaya bloom, River plume Off Mississippi Delta ( N, W) Senescent diatom bloom ( N, W) Mar 93 Apr 93 Apr 92 Jun 94 Ju193 Ju193 May Incuba- DOC in various tion size fractions time* (h) DOC LMW HMW VHMW responding LMW components. The present study extends these investigations to waters of varying DOM source and composition, including DOM from riverine, estuarine, and marine environments. In order to directly compare the bioreactivity of different molecular size fractions of DOM, HMW (> 1 kda) and LMW fractions of natural DOM were separated by tangential-flow ultrafiltration and incubated under similar conditions with a natural bacterial assemblage. Bacterial growth and respiration were measured to directly compare the potential bioreactivity of the LMW and HMW fractions. The study was designed to determine the rates and extent of bacterial degradation of natural DOM size fractions from various environments during incubations lasting 3-5 d. The results of these investigations provide additional insight to the relationships between reactivity and origin of different size classes of DOM in distinct environments. A new conceptual model for DOM degradation- the size-reactivity continuumis introduced to account for the high potential reactivity of HMW DOM. Material and methods Study sites were chosen to represent a wide spectrum of DOM sources and compositions. Sampling sites in the Amazon were chosen - 3 km upstream of the confluence of the Rio Negro and the Rio Solimoes. Both rivers derive their DOM almost exclusively from terrestrial vegetation (Quay et al. 1992; Hedges et al. 1994). Phytoplankton production is limited in both rivers due to turbidity and humic coloration. Chemical analyses of DOM indicates extensive diagenetic alteration in both rivers (Hedges et al. 1994). The Laguna Madre is at the southern end of the Texas Gulf Coast and stretches 200 km from Corpus Christi to the Rio Grande. The Laguna Madre is a shallow subtropical lagoon with abundant and highly productive seagrass meadows. Opsahl and Benner (1993) studied the decomposition of seagrass detritus in the lagoon and found that a major fraction of the material is rapidly solubilized and enters the dissolved pool. The northern Gulf of Mexico receives considerable quantities of terrestrially derived organic matter from the Mississippi River system. The river plume also carries high concentrations of inorganic nutrients that stimulate phytoplankton blooms at midsalinities (Lohrenz et al. 1990). DOM in this region is largely terrestrially derived at low salinities and marinederived at higher salinities (Benner et al. 1992a). The Aransas Pass channel is in Port Aransas, Texas, and was sampled from the Marine Science Institute (University of Texas) pier. The channel carries a mixture of estuarineand marine-derived DOM. Sampling and experimental procedures - Sampling was conducted at the above sites under various hydrological and biological conditions (Table 1). During two cruises in May 1992 and July 1993, experiments were conducted on the RV Longhorn off the Mississippi Delta in the northern Gulf of Mexico (GOM). Samples were taken with a clean bucket from the surface and immediately processed as described below. One 50-liter sample was taken in the upper Laguna Madre. The sample was collected in the morning, transported to the laboratory, and processed within 1 h of sampling. During 2 months in Manaus, Brazil, experiments were performed with water collected from the Rio Negro and the Rio Solimoes. Samples were taken with a bucket from a small boat and transported to the laboratory. Samples were processed within 2 h of sampling. The initial water sample (- 25 liters) was passed through a 0.6-pm pore-size Nuclepore polycarbonate cartridge filter to remove particulate material and most eucaryotes (Fig. 1). The filtrate was then passed through a hollowfiber filter (0.1-pm pore-size) using a tangential-flow ultrafiltration system (Amicon DC 1 OL) to concentrate bacterial cells for later use as an inoculum (Benner 1991). In all experiments, we used tangential-flow ultrafiltration and spiral-wound ultrafilters to separate HMW DOM (>30 kda or > 1 kda) from LMW DOM (< 1 kda). In most experiments, HMW refers to DOM > 1 kda. In the two

3 Bacterial utilization of DOM 43 Amazon River experiments, we used an additional filter (30 kda) and investigated the reactivity of three size fractions, very-high-molecular-weight (VHMW, > 30 kda), HMW (~30 kda, > 1 kda) and LMW (< 1 kda) DOM. This method and Amicon ultrafilters have been shown to be noncontaminating and to allow rapid sample processing (Benner 1991; Benner and Hedges 1993). A concentration factor of 10 was used for the separation of HMW and LMW DOM. After size separation, the concentrate (> 30 kda or > 1 kda) was diluted with lowcarbon distilled water or artificial seawater to a similar DOC concentration as expected in the LMW fraction. Parallel treatments with DOM of varying molecular weight received equal additions of inorganic nutrients. Nitrate (KNO,) was chosen over ammonia as a nitrogen source to avoid potential stimulation of nitrification and interference with respiration measurements. Nitrate and phosphate (NaH,PO,) additions were -3-fold greater than ambient concentrations. Equal portions of the bacterial concentrate isolated from each water sample were added to all parallel treatments. Water samples were placed in acid-washed glass bottles (Fig. 2). The bottles were filled without a headspace, sealed with silicon stoppers, and connected by Teflon tubing. Each incubation consisted of two connected bottles and was conducted in duplicate. Sampling was performed by injecting water (same treatment) into the reservoir bottle and simultaneously sampling from the other bottle (see Fig. 2). During an entire experiment, < 7% of the volume was exchanged in the sampling bottle. Samples were withdrawn over a period of 3-5 d at h intervals. Measurements-Bacterial abundance was determined by epifluorescence microscopy using DAPI as a stain (Porter and Feig 1980). Marine samples were preserved with formaldehyde and freshwater samples with boraxbuffered formaldehyde (4% final concn). River samples were diluted (1: 10 with 0.2-pm-filtered sample water), and a higher concentration of the stain (1 pg DAPI ml- l) was used. At least 300 cells were counted per sample. Bacterial production was estimated from rates of protein synthesis as measured from rates of radiolabeled leucine incorporation (Kirchman et al. 1985). Triplicate water samples (10 ml) and one killed control were incubated with 10 nm (final concn) of [4,5-3H] leucine (sp act, 2.22 GBq hmol-l) for 0,5 h in the dark and at ambient temperature. Leucine incorporation values were converted to cells produced by means of empirically derived conversion factors and a cell-to-carbon conversion value of 20 fg C cell- l (Lee and Fuhrman 1987). Leucine conversion factors for the different environments were 4.27 x 1016 cells mol-l for the Laguna Madre (Chin-Leo and Benner 199 l), 4.46 x 1016 cells mol-l for the Gulf of Mexico (Chin- Leo and Benner 1992), and 0.74 x 1016 cells mol-l for the Amazon River system. Empirically derived factors were chosen to account for the variability among environments and because bottle incubations represent conditions similar to those used for conversion factor determinations. DOC concentrations were determined by high-temperature combustion after sample filtration through pre- Bacterial concentrate + (500 ml) -2.5 liters of concentrate t t Dilution with ASW or DW (-15 liters) t +N&P (3X natural concnl Water sample (25 liters) Nuclepore polycarbonate filter (0.6 pm) Hollow-fiber filter (0.1 Pm) Filtrate: -24 liters (CO. 1 pm) Spiral filter (1 kda or 30 kda) * -21 liters of filtrate t +N&P (3X natural concn) t ml of bacterial ml of bacterial concentrate concentrate Fig. 1. Protocol for preparing water samples for the bioreactivity experiments. Large particles and most eucaryotic organisms were removed by passage through a 0.6~pm pore-size Nuclepore cartridge filter. The sample was then passed through a hollow-fiber filter (0.1 -vm pore-size) by tangential-flow ultrafiltration, and 500 ml of concentrate was kept as a bacterial inoculum. Tangential-flow ultrafiltration with spiral ultrafilters (1 or 30 kda mol wt cutoff) was used for molecular size separation of the dissolved organic matter (DOM). After ultrafiltration, the concentrate was diluted with low-carbon artificial seawater or distilled water. Nitrate, phosphate (3 x natural concn), and 100 ml of the bacterial cell concentrate from the hollowfiber filtration were added to both the diluted concentrate and the filtrate. ASW - artificial seawater; DW - distilled water; N - nitrate; P-phosphate. combusted GF/F filters (Benner and Strom 1993). The bacterial cells that passed through this filter were included in the DOC measurement. Oxygen concentrations were determined by the Winkler method with a Mettler DL 2 1 autotitrator with potentiometric end-point detection (Grankli and Grankli 199 1). Samples for oxygen determination were collected in acid-washed and rinsed BOD bottles (60 ml). Respiration rates were calculated between the initial and the peak-abundance time point of the bacterial population as Oi - Of r=at r is the respiration rate, Oi, is the initial oxygen concen- (1)

4 44 Amon and Benner Injection of 120 ml of incubation medium Teflon valve Fig. 2. Experimental setup showing incubation bottles and sampling procedure. Arrow indicates dire&ion of flow during sampling. tration, Qf is the oxygen concentration at the peak of bacterial abundance, and At is the incubation time (h). The value for t was variable among the experiments and is listed in Table 1. Bacterial growth efficiencies in the different size fractions of DOM were calculated from bacterial C production (estimated either from leucine incorporation or changes in bacterial abundance) and bacterial respiration, assuming a respiratory quotient (RQ) of 1. Ecfect of substrate concentration on rates of utilization - An additional experiment was performed to determine the influence of DOC concentration on bacterial DOC utilization rates. A water sample (280/,) was taken from Aransas Pass at the Marine Science Institute pier, and molecular size separation was performed as described above. The HMW concentrate was diluted to three different concentrations. One HMW incubation had a lower DOC concentration than the LMW fraction, one had a concentration equal to the LMW fraction, and one had a higher concentration. Results The DOC concentrations were highest ( PM DOC) in the rivers and low-salinity samples and were fold the concentration found in the GOM sample at 27?& (152 PM; Table 1). Attempts to balance the initial DOC concentrations in the different experimental treatments (VHMW, HMW, LMW) were not successful because we did not have access to the DOC analyzer when these experiments were initiated in the field. The DOC concentrations in the LMW fractions were typically fold lower than concentrations in the respective HMW fractions (Table 1). The fraction of total DOC in the HMW and LMW fractions varied considerably among sites (Fig. 3). In general, the percentage of DOC in the HMW fraction decreased with increasing salinity. Most DOC in the Rio Negro and Rio Solimdes was in the HMW fraction (88 and 8 l%, respectively). In the low-salinity GOM samples and the Lag una Madre, - 60% of the DOC was recovered in the HMW fraction. In contrast, higher salinity samples from the GOM had a greater proportion of the DOC in the LMW fraction. About 70% of the DOC in the 27%0 sample from the GOM was in the LMW fraction. The Aransas Pass sample had a mixture of estuarine and oceanic DOM that resulted in about equal amounts of HMW and LMW DOC. In some of the experiments, especially in the river and low-salinity samples, the precision of the DOC measurements was insufficient to detect statistically significant changes in DOC concentrations. Direct measurements of DOC utilization at 27?& were presented earlier (Amon and Benner 1994) and were also detected in some of the experiments presented herein. However, to be consistent among the different experiments presented in this study, we used bacterial abundance, leucine incorporation, and oxygen consumption as measures of bacterial utilization of different molecular size fractions of DOC. All rates presented he,re were calculated for the period of increasing bacterial abundance (from the initial value to the peak abundance t,alue), which differed among experiments (see Table I). Typical time-course data are presented for two representative experiments, and the rates of bacterial growth and remineralization are summarized in the text and in tables. The sample taken from the Atchafalaya River plume in the northern GOM at 8%0 was dominated by riverine DOM, but this water also contains phytoplankton-derived DOM supplied by an ongoing cyanobacteria bloom during sampling. Oxygen concentrations show a much

5 Bacterial utilization of DOM I Rio 1 Negro Rio Solim oes Laguna Madre GOM COM ! I I I I I I B Salinity (%o) Fig. 3. Molecular weight distribution of DOM in different environments as determined by tangential-flow ultrafiltration (concn factor of 10). Relative amounts are given as percentages of total DOC. Dark bars indicate the HMW fraction (> 1 kda) and light bars the LMW fraction (< 1 kda); Gulf of Mexico- GOM c c greater decline in the HMW than in the LMW fraction (Fig. 4A). Bacterial leucine incorporation rates (Fig. 4B) were consistently higher in the HMW fraction, with a peak value of 0.88 nm leucine h- l as compared to 0.53 nm leucine h-l in the LMW fraction. The same trend was observed for bacterial abundances (Fig. 4C), with peaks of 4.50 x 1 Og cells liter -I for the HMW fraction and 3.20 x 1 Og cells liter -l for the LMW fraction. In the experiment with GOM water (23Ym), phytoplankton-derived marine DOM became quantitatively more important than riverine DOM. Again, we observed the same trend of higher reactivity in the HMW fraction than in the LMW fraction (Fig. 5). More oxygen was consumed in the HMW fraction than in the LMW fraction during the incubation (Fig. 5A). The respiration rate was 4 times higher in the HMW (0.53 PM O2 h-l) than in the LMW fraction. For comparison, the in situ respiration rate at the sampling site was 0.95 PM O2 h-l (Pakulski et al. in prep.). Leucine incorporation was consistently higher in the HMW fraction, with a peak rate of 6.20 nm leucine h- l (Fig. 5B). Peak bacterial abundance values were 2.30 x log cells liter-l in the HMW and 1.92 x log cells liter- in the LMW fraction (Fig. 5C). The experiment to test the influence of DOC concentration on rates of bacterial remineralization and growth demonstrates that at the same initial DOC concentration, the HMW fraction stimulates higher rates of bacterial growth and respiration than the LMW fraction does (Fig. 6). The respiration rate in the HMW fraction was nearly twice as high as the rate in the LMW fraction, and even at lower initial DOC concentrations in the HMW fraction, we observed higher respiration rates than in the LMW 1 I I I I I t Time (h) Fig. 4. Bacterial oxygen consumption, bacterial leucine incorporation, and bacterial abundance in incubations with natural DOM of different molecular weights from Atchafalaya River plume water (8%). Plume water with HMW DOM (> 1 kda)- 0; plume water with LMW DOM (< 1 kda)-a. Each point represents the average of two replicate experiments; error bars represent their mean deviation. fraction (Fig. 6A). The C-normalized amount of oxygen consumed for the three different concentrations (116,167, and 302 PM DOC) of HMW DOC were similar (106.7, 93.6, and 94.1 PM 0, mmol DOC- ), while the C-normalized value for the LMW DOM was much lower (57.36 PM O2 mmol DOC- ), indicating that HMW DOM was clearly more bioreactive than was LMW DOM. Bacterial leucine incorporation rates were also higher in the HMW fraction (3.89 nm leucine h- l) than these in the LMW (3.2 1 nm leucine h-l) fraction (Fig. 6B). This experiment also demonstrated the strong dependence of respiration rates and bacterial growth rates on DOC concentration.

6 46 Amon and Benner A I I I I I I I - 8 rl 2 B I I 0 1; ;4 BO T C A 3.0-I I I I I I I I I t I I I I I I Time (h) Fig. 5. As Fig. 4, but from the Gulf of Mexico (23o/oo). PM DOC Fig. 6. Rates of bacterial respiration and bacterial leucine incorporation in Aransas Pass water (28%~) with three different initial HMW DOC concentrations and a single concentration of LMW DOC. HMW DOM (>l kda)-0; LMW DOM (~1 kda-a. The effects of DOC concentrations on rates of bacterial growth and respiration are compared with effects of molecular size. The relationship between DOC concentration in the HMW fraction and rates of bacterial respiration and growth was highly significant (r = 0.997, P < for respiration; r = , P < 0.05 for bacterial leucine incorporation), indicating that comparisons between utilization rates of HMW and LMW DOM should be made on a C-normalized basis. Therefore, all rates of bacterial activity are presented as C-normalized values in Table 2. Overall, rates of bacterial cell production in marine samples were higher than rates in river samples (Table 2). The VHMW fraction supported the highest rate of bacterial cell production in the Rio Negro [34.8 x 1 O6 cells liter- h-l (mmol DOC)- ]; highest rates in the Rio Solimoes were observed in the HMW fraction [46.7 x lo6 cells liter-l h-l (mmol DOC)- ]. In most other experiments (except Laguna Madre and GOM 23Y&~), the HMW fraction yielded higher rates of bacterial cell production than did the LMW fraction. A very high rate [99.8 x lo6 cells liter- 1 h- 1 (mmol DOC)- 1 of cell production was observed in the HMW fraction of the diatom bloom sample at 27o/oo in the GOM. Time-integrated leucine incorporation rates (Table 2) followed the same overall trend as the cell production measurements, with lowest overall rates in river and lowsalinity samples. In all but one experiment, HMW DOM stimulated higher rates of leucine incorporation than did LMW DOM. In the Rio Negro, the VHMW DOM showed a higher rate [8.28 nm leucine h-l (mmol DOC)- :] than was shown by the other two size fractions, and in the Rio Solimdes, HMW DOM stimulated a rate higher [8.52 nm leucine h-l (mmol DOC)- 1 than that of VHMW or LMW DOM. The highest rate of leucine incorporation was observed with HMW DOM from the diatom bloom sample at 27o/oo.

7 Bacterial utilization of DOM 47 Table 2. Bacterial growth [cell production, lo6 cells liter-l h-l (mmol DOC)-l] and respiration [PM 0, h-l (mmol DOC)- ] of different molecular-weight fractions of DOM from several aquatic environments. All rates were calculated for the growth phase of the bacterial population as given in Table 1. Leucine incorporation rates [nm leucine h-l (mmol DOC)) ] were integrated over the growth phase. DOC utilization rates [PM DOC h-l (mmol DOC)-l] represent the sum of respiration (assuming an RQ of 1) and bacterial production (estimated from leucine incorporation). All rates are normalized to the initial DOC concentration in the incubation. Gulf of Mexico - GOM. Cell production Respiration rate Leu incorporation DOC utilization Location VHMW HMW LMW VHMW HMW LMW VHMW HMW LMW VHMW HMW LMW Rio Negro Rio Solimoes * Laguna Madre * Aransas pass GOM 8Ym , 1.09 GOM 23% GOM 27Yoo * ANOVA indicated that the slope of a linear regression through the data points was not significantly different from zero (P = 0.74 for the Rio Solimoes; 1 = 0.99 for the Laguna Madre). C-normalized respiration rates (Table 2) also were highest in HMW DOM from the diatom bloom at 27Ym [7.08 PM O2 h-l (mmol DOC)- ]. Respiration rates were lowest in the experiments with LMW DOM from the Rio Solimoes [O. 12 PM O2 h-l (mmol DOC)- 1. Respiration rates for the Rio Negro and Rio Solimoes samples tracked measurements of bacterial growth, with highest respiration rates in the VHMW fraction from the Rio Negro and the HMW fraction from the Rio Solimoes. In all experiments, HMW DOC (VHMW and HMW in rivers) was remineralized at a rate higher than that of LMW DOC. C-normalized respiration rates were typically 2-4 times higher in the HMW fractions. C-normalized rates of DOC utilization include remineralization and bacterial biomass production and are therefore the most complete measure of the bioreactivity of different molecular size fractions of DOC. The DOC utilization rates in Table 2 were calculated from respiration rates and bacterial carbon production rates estimated from leucine incorporation rates. From these results, we conclude that HMW and VHMW DOC were utilized at a similar or a higher rate than LMW DOC in all the environments investigated. These rates are normalized for the different initial DOC concentrations, suggesting that there are qualitative distinctions between HMW and LMW substrates. The fraction of DOC utilized during incubations with the different molecular size fractions (VHMW, HMW, LMW) are summarized in Table 3 and reflect the proportion of total DOC that is readily available for bacterial uptake. The fraction of DOC utilized is normalized for time (% of DOC utilized d-l) because of the varying lengths of the growth phases in the different experiments (Table 1) and was calculated by the formula C r-p + %rod(100). F = ~DWfract (2) F is the percentage of DOC utilized per day, Cresp is the amount of carbon respired per day as calculated from oxygen consumption assuming an RQ of 1, Cprod is the amount of carbon produced by bacteria per day,as calculated from leucine incorporation rates (see methods), and [DOCL is the initial DOC concentration in the corresponding molecular size fraction. This calculation allows us to compare the bioavailability of DOM in the different environments as well as the different molecular size fractions. In all experiments, the fraction ( % d-l) of HMW DOM utilized was greater than that of LMW DOM ( % d-l). The percentage ( % d-l) of the DOC in the various size fractions utilized in the river samples was lower than that in marine samples ( % d-l). This trend indicates that terrestrially derived DOM in rivers is less bioavailable than is phytoplankton-derived DOM. Moreover, the very high percentage of DOC utilized in the diatom bloom sample indicates that freshly produced DOM is the most bioavailable. Bacterial growth efficiencies were calculated from respiration and bacterial production as given in Eq. 3: Table 3. Relative bioreactivities of DOM of varying molecular sizes and bacterial growth efficiencies (GE) on DOM from different environments. All values were calculated with the formulas given in the text. Gulf of Mexico-GOM. % DOC utilized (d-9 GE (%) Location VHMW HMW LMW VHMW HMW LMW Rio Negro Rio Solimoes Laguna Madre Aransas Pass GOM 8?& GOM 23% GOM 27o/oo* * In an earlier publication (Amon and Benner 1994), the value for the fraction of DOC consumed differed from that presented here because utilization rates (bacterial respiration plus production) were used rather than respiration rates alone.

8 48 Amon and Benner GE = &m GE is growth efficiency (o/o), P is bacterial production in carbon units as calculated from leucine incorporation rates, and R is bacterial respiration in carbon units calculated from dissolved oxygen consumption (RQ = 1) for the incubation period given in Table 1. A consistent pattern throughout all experiments was observed, with GEs in the LMW incubations higher than those in HMW incubations. Bacterial GEs ranged from 8-39% in the HMW fractions and 14-66% in the LMW fractions (Table 3). In most cases, the difference in GE values on the various size fractions of DOC is pronounced; only the Rio Negro showed similar GEs for the three size fractions. Discussion We investigated the bacterial utilization of different molecular size fractions of DOC to measure their relative bioavailabilities and to relate bioavailability to molecular size and diagenetic state. The experiments revealed that across a spectrum of environments with DOM of varying sources and compositions, the reactive pool of HMW DOC is typically larger than the reactive pool of LMW DOC. This finding was surprising and calls for a modified conceptual model for DOM decomposition. Traditional models, such as the one presented by Saunders (1976), state that simple molecules (LMW) decompose faster than HMW compounds do. The same conceptual model was recently presented by Miinster and Chrbst (1990), who stated that organic molecules with molecular weight < 1 kda are preferentially used by natural bacterioplankton and that the rate of utilization of phytoplankton-derived DOM decreases with increasing molecular weight. Advances in Chromatographic separation techniques and increases in analytical sensitivity have stimulated many studies of the bacterial utilization of LMW compounds such as dissolved free amino acids (Fuhrman 1987; Keil and Kirchman 1991; Suttle et al. 1991). These studies have demonstrated that some LMW components cycle rapidly and contribute to bacterial growth, but these compounds comprise a relatively small fraction of natural DOM. The origin of this model goes back to the early 1960s when researchers observed high rates of utilization of simple radiolabeled substrates added to natural water samples (Wright and Hobbie 1965). Our results indicate that the rapid utilization rates observed with added LMW compounds are not applicable to the bulk of naturally occurring LMW DOM. In contrast to the traditional model, our results indicate that along a continuum of molecular sizes, DOM seems to decrease in bioreactivity with decreasing size. We believe this relationship exists because the origin of HMW components tends to be more recent than that of most LMW components. Molecular size seems to be a general indicator for the diagenetic state of DOM, with decreasing molecular size as diagenetic alteration progresses. Studies of the chemical composition of organic matter of varying sizes and {diagenetic states in natural environments indicate that a greater fraction of HMW components than of LMW components can be chemically characterized (Hama and Handa 1980). The fraction of fresh phytoplankton detritus that can be chemically characterized is also greater than the fraction of diagenetically altered POM that can be chemically characterized (Lee and Wakeham 1988). A recent study of the amino acid and carbohydrate compositions of dissolved and particulate organic matter in the Amazon River revealed that the three size classes of organic matter investigated- coarse particulate (CPOM), line particulate (FPOM), and ultrafiltered dissolved organic matter (> 1 kda; UDOM)- were chemically distinct (Hedges et al. 1994). The UDOM fraction was equivalent to the HMW plus VHMW DOM fractions in the present study. The CPOM most closely resembled its vascular plant source, FPOM represented more highly degraded material, and the UDOM fraction was the most profoundly degraded. These results are consistent with a size-reactivity continuum model and extend it from dissolved to particulate size classes of organic matter. In a similar fashion, diagenetic indicators have also been applied to phytoplankton-derived material in order to follow diagenetic changes. Fresh phytoplankton, the main source of organic matter in the ocean, was compared to organic matter collected in sediment traps and to organic matter in sediments (Lee and Cronin 1982; Cowie and Hedges 1994). These studies indicate that fresh phytoplankton. material undergoes pronounced diagenetic alteration during degradation as it sinks through the water column. 0 nly a few percent of the source material reaches the sediment surface (Lee and Wakeham 1988), indicating that the major portion of POM is either remineralized directly or transformed to DOM. The results of our study of the bioreactivity of HMW and LMW DOM and evidence from studies of the diagenetic stai;e of organic matter in varying size classes support the size-reactivity continuum model. In a general way, diagenetically young or fresh material seems to be most bioreactive. During decomposition, organic matter continuously becomes less bioreactive and smaller in physical size. A schematic flow diagram for the size-reactivity continuum model is presented in Fig. 7; it indicates that the m(ajor pathway of organic matter degradation is a continuous change from bioreactive organic particles and macromolecules to small refractory organic molecules. It is important to recognize that each size fraction (POM, HMW, and LMW) consists of a variety of compounds wil;h distinct compositions, reactivities, and diagenetic states. The size-reactivity continuum model and differences in the bioreactivity of the various size classes of DOM seem to be, reflected in the overall concentrations of organic matter in each size fraction observed in the different environments (Fig. 3). Most DOM in marine samples (60-70%) passed the 1 kda filter, whereas >80% of the DOM was retained by the same filter in Amazon River samples. These molecular weight distributions agree with previous wudies, indicating that most riverine DOM is

9 Bacterial utilization of DOM 49 of HMW (Meyer et al. 1987; Benner and Hedges 1993; Hedges et al. 1994), whereas most DOM in the ocean is of LMW (Carlson 1985; Benner et al. 1992b; Ogawa and Ogura 1992). Furthermore, the molecular size of oceanic DOM decreases with increasing depth (Benner et al. 1992b; Ogawa and Ogura 1992). The bulk of LMW DOC represents highly degraded material and is the least reactive fraction of DOM and, as such, accumulates in the deep ocean. Very recently, 14C age determinations on different molecular-size fractions of marine DOM showed that colloidal (> 10 kda) DOM is younger (few decades) than DOM of lower molecular weight (380-4,500 yr; Santschi et al. 1995). Riverine environments can be considered a transition zone and are typically dominated by diagenetically younger (< 150 yr; Hedges et al. 1994) HMW DOM. Combined information on the molecular size distribution of DOM in different environments and 14C age determinations on DOM represent a time-integrated measure for DOM reactivity and are consistent with the size-reaci tivity continuum model. The influence of the diagenetic state of organic matter on its bioreactivity is apparent when rates of utilization of DOC fractions from different environments are compared. HMW DOC from the Amazon is clearly less reactive than is HMW DOC collected from a phytoplankton bloom (277~). The HMW DOC from the Amazon has undergone extensive degradation (Hedges et al. 1994), whereas the HMW DOM from the phytoplankton bloom is of more recent origin. Our experiments revealed several findings suggestive of compositional or qualitative differences among the various size classes of DOM. Variable bacterial growth efficiencies indicate that the reactive substrates in the HMW and LMW fractions are compositionally distinct. In all of the investigated environments, we observed higher bacterial growth efficiencies in the LMW fraction. Leucine incorporation was used as a measure of bacterial production in calculating the bacterial growth efficiencies presented in Table 3. If bacterial abundance values are used to estimate growth efficiencies in this study, the overall growth efficiencies would be lower by a factor of -4 (data not shown). However, the trend would be the same, with growth efficiencies in the LMW higher than in the HMW fractions. The higher growth efficiencies in the LMW fraction indicate that the available LMW substrates are richer in organic N and other bioactive elements required for bacterial growth (Goldman et al. 1987). This suggests that the available portion of LMW DOM has a lower C : N ratio than that of the reactive fraction of the HMW material. In the GOM experiment at 279~, nutrient concentrations were directly measured. Inorganic nutrient dynamics in the HMW and LMW incubations were dramatically different (Amon and-benner 1994). Inorganic N was consumed during bacterial degradation of the HMW fraction and regenerated in the LMW fraction, indicating that bioreactive HMW material is depleted in organic N. This study also demonstrates that low growth efficiencies do not necessarily indicate that a certain DOC fraction is less bioreactive. Size-reactivity continuum model High 4 REAC IIVI I-Y * LOW Fig. 7. Schematic diagram of the size-reactivity continuum model for organic matter decomposition in aquatic environments. Arrow indicates the major pathway of degradation from bioreactive organic particles and macromolecules to refractory LMW compounds. The size of the dots is representative of the size of organic matter, with large dots for particulate organic matter, medium-sized dots for HMW DOM, and small dots for LMW DOM. The distribution of dots indicates that most larger sized organic matter is more reactive than most smaller sized organic matter. Relatively high C : N ratios were directly measured in oceanic as well as Amazon River HMW DOM (Benner et al. 1992b; Hedges et al. 1994). The reactive components of LMW DOM seem to be rich in organic N (low C : N ratio), suggesting that dissolved free amino acids may be important bioreactive components of the LMW fraction. These C and N dynamics are strongly supported by a recent study of the effect of HMW DOM and light on heterotrophic nitrogen dynamics (Gardner et al. in prep). That study demonstrated rapid bacterial uptake of ammonia and amino acids in the presence of marine HMW DOM; ammonia was regenerated by bacteria in the presence of LMW DOM. In our experiments, we added nitrate rather than ammonium as a nitrogen source to avoid potential stimulation of nitrification. Bacterial utilization of nitrate during the utilization of HMW DOM could explain the relatively low values of bacterial GE in the HMW fractions due to the high energy demand associated with nitrate assimilation. Nitrate has been viewed as a minor source of N for bacteria, but recent studies (Kirchman et al. 1991, 1994) indicate that bacteria can utilize a considerable amount of nitrate under certain conditions and that nitrate can support up to 10% of bacterial N production in the ocean (Kirchman et al. 1994). The ability of bacteria to grow on the HMW fraction of natural DOM has been reported from several environments. Most studies of this kind were conducted in freshwaters rich in humic substances (Meyer et al. 1987; Tranvik 1990; Tulonen et al. 1992). Billen et al. (1990) suggested the importance of macromolecular compounds

10 50 Amon and Benner for bacterial growth on phytoplankton-derived DOM in the marine environment, and Arnosti et al. (1994) recently reported rapid bacterial decomposition of polysaccharides in anoxic marine systems. The size-reactivity continuum model (Fig. 7) suggests that degradation of organic matter in aquatic environments leads to formation of refractory LMW compounds. Hatcher and Spiker (1988) presented a general scheme for the humification process that is consistent with the size-reactivity continuum model. In that scheme, labile macromolecules are rapidly utilized during early diagenesis; refractory compounds are selectively preserved and become what is operationally defined as humin. Further degradation leads to oxidation of humin and formation of more soluble humic and fulvic acids. Overall, humification leads to increasingly oxidized substrates of lower molecular weight and greater solubility. This suggests that refractory LMW compounds in the ocean no longer resemble biomolecules and thereby escape microbial utilization. Additional studies of the composition and reactivity of DOM size fractions are required to further evaluate the proposed size-reactivity continuum model and to better understand the significance of refractory LMW DOM in the global carbon cycle. References AMON, R. M. W., AND R. BENNER Rapid cycling of high-molecular-weight dissolved organic matter in the ocean. Nature 369: ARNOSTI, C., D. J. REPETA, AND N. V. BLOUGH Rapid bacterial degradation of polysaccharides in anoxic marine systems. Geochim. Cosmochim. Acta 58: BAUER, J. E., P. M. WILLIAMS, AND E. R. M. DRUFFEL C activity of dissolved organic carbon fractions in the north-central Pacific and Sargasso Sea. Nature 357: BENNER, R Ultrafiltration for the concentration of bacteria, viruses, and dissolved organic matter, p. 18 l-l 85. In D. C. Hurt and D. W. Spencer [eds.], Marine particles: Analysis and characterization. -, C. CHIN-LEO, W. GARDNER, B. EADIE, AND J. COTNER. 1992~. The fates and effects of riverine and shelf-derived DOM on Mississippi River plume/gulf shelf processes, p In Nutrient enhanced coastal ocean productivity. NECP Workshop Proc. NOAA Coastal Ocean Program. TAMU-SG , AND J. I. HEDGES A test of the accuracy of freshwater DOC measurements by high-temperature catalytic oxidation and UV-promoted persulfate oxidation. Mar. Chem. 41: , J. D. PAKULSIU, M. MCCARTHY, J. I. HEDGES, AND P. G. HA rcher. 1992b. Bulk chemical characteristics of dissolved organic matter in the ocean. Science 255: , AND M. STROM A critical evaluation of the analytical blank associated with DOC measurements by high-temperature catalytic oxidation. Mar. Chem. 41: BILLEN, G., AND OTHERS Role of bacteria in the North Sea. Neth. J. Sea Res. 26: CARLSON, D. J Molecular weight distribution of dis- solved organic materials in seawater determined by ultrafiltration: A re-examination. Mar. Chem. 16: 155-l 7 1. CHIN-LEO, G., AND R. BENNER Dynamics of bacterioplankton abundance and production in seagrass communities of a hypersaline lagoon. Mar. Ecol. Prog. Ser. 73: , ANCl Enhanced bacterioplankton production and respiration at intermediate salinities in the Mississippi River plume. Mar. Ecol. Prog. Ser. 87: COWIE, G.,,QND J. HEDGES Biochemical indicators of diagenetic alteration in natural organic matter mixtures. Nature 369: FARRINGTON, J Overview and key recommendations. Marine organic geochemistry workshop, January Mar. Chem. 39: 5-9. FUHRMAN, J Close coupling between release and uptake of dissolved free amino acids in seawater studied by an isotope dilution approach. Mar. Ecol. Prog. Ser. 37: GOLDMAN, Ji. C., D. A. CARON, AND M. R. DENNETT Regulation of gross growth efficiency and ammonium regeneration in bacteria by substrate C:N ratio. Limnol. Oceanogr. 32: GRANI?LI, W., AND E. GRANBLI Automatic potentiometric determination of dissolved oxygen. Mar. Biol. 108: s. HAMA, T., AND N. HANDA Molecular weight distribution and characterization of dissolved organic matter from lake waters. Arch. Hydrobiol. 90: 106-l 20. HATCHER, P., AND E. SPIKER Selective degradation of plant biomolecules, p In F. Frimmel and R. Christman [eds.], Humic substances and their role in the environmen t. Wiley. HEDGES, J. I., AND OTHERS Origins and processing of organic matter in the Amazon River as indicated by carbohydrates and amino acids. Limnol. Oceanogr. 39: &IL, R. G., AND D. L. KIRCHMAN Contribution of dissolved amino acids and ammonium to the nitrogen requirements of heterotrophic bacterioplankton. Mar. Ecol. Prog. Ser. 73: l-10. KIRCHMAN, D. L., H. W. DUCKLOW, J. J. MCCARTHY, AND C. GARSIDIZ Biomass and nitrogen uptake by heterotrophic bacteria during the spring phytoplankton bloom in the North Atlantic Ocean. Deep-Sea Res. 41: , E. K. NEES, AND R. HODSON Leucine incorporation and its potential as a measure of protein synthesis by bacteria in natural aquatic systems. Appl. Environ. Microbiol. 49: , Y. SUZUKI, C. GARSIDE, AND H. W. DUCKLOW High turnover rates of dissolved organic carbon during a spring phytoplankton bloom. Nature 352: LEE, C., AND C. CRONIN The vertical flux of particulate organic nitrogen in the sea: Decomposition of amino acids in the Peru upwelling area and the equatorial Atlantic. J. Mar. Res. 40: AND S. G. WAKEHAM Organic matter in seawater: Eliogeochemical processes, p. l-5 1. In J. Riley [ed.], Chemical oceanography. V. 9. Academic. LEE, S. AND J. A. FUHRMAN Relationships between biovolume and biomass of naturally derived marine bacterioplankton. Appl. Environ. Microbial. 53: 1298-l 303. LOHRENZ, S, E., M. J. DAGG, AND T. E. WHITLEDGE Enhanced primary production at the plume/oceanic interface oft he Mississippi River. Cont. Shelf Res. 10: MEYER, J. L., R. T. EDWARDS, AND R. RISLEY Bacterial

11 Bacterial utilization of DOM 51 J growth on dissolved organic carbon from a blackwater river. Microb. Ecol. 13: MUENSTER, U., AND R. J. CHRIST Origin, composition and microbial utilization of dissolved organic matter, p In J. Overbeck and R. J. Chrost [eds.], Aquatic microbial ecology. Biochemical and molecular approaches. Springer. OGAWA, H., AND N. OGURA Comparison of two methods for measuring dissolved organic carbon in sea water. Nature 356: OPSAHL, S., AND R. BENNER Decomposition of senescent blades of the seagrass Halodule wrightii in a subtropical lagoon. Mar. Ecol. Prog. Ser. 94: 19 l-205. POMEROY, L. R The ocean s food web, a changing paradigm. Bioscience 24: PORTER, K. G., AND Y. S. FEIG The use of DAPI for identifying and counting aquatic microflora. Limnol. Oceanogr. 25: QUAY, P. D., AND OTHERS Carbon cycling in the Amazon River: Implications for the 13C composition of particles and solutes. Limnol. Oceanogr. 37: SANTSCHI, P. H., AND OTHERS Isotopic evidence for the contemporary origin of high-molecular weight organic matter in oceanic environments. Geochim. Cosmochim. Acta 59: SAUNDERS, G Decomposition in fresh water, p In J. Anderson and A. Macfadyen [eds.], The role of terrestrial and aquatic organisms in decomposition processes. Blackwell. SUTTLE, C. A., A. M. CHAN, AND J. A. FUHRMAN Dissolved free amino acids in the Sargasso sea: Uptake and respiration rates, turnover times, and concentrations. Mar. Ecol. Prog. Ser. 70: TRANVIK, L. J Bacterioplankton growth on fractions of dissolved organic carbon of different molecular weights from humic and clear waters. Appl. Environ. Microbial. 56: TULONEN, T., K. SALONEN, AND L. ARVOLA Effects of different molecular weight fractions of dissolved organic matter on the growth of bacteria, algae and protozoa from highly humic lakes. Hydrobiologia 229: WILLIAMS, P. M., AND E. R. M. DRUFFEL Radiocarbon in dissolved organic matter in the North Pacific Ocean. Nature 330: WRIGHT, R., AND J. HOBBIE The uptake of organic solutes in lake water. Limnol. Oceanogr. 10: Submitted: 14 April I995 Accepted: 30 August 1995 Amended: 25 October 1995