BENNER, RONALD, JAN LAY, ELIZABETH K'NEES, AND ROBERT E ...

19 downloads 53 Views 1MB Size Report
Jan Lay, Elizabeth K'nees, and Robert E. Hodson. Department of Microbiology ...... ation than previously thought (Goldman and. Caron 1985; Sherr et al. 1986 ...
Limnol. Oceanogr., 33(6, part 2), 1988, 1514-1526 0 1988, by the American Society of Limnology and Oceanography, Inc.

Carbon conversion efficiency for bacterial growth on lignocellulose: Implications for detritus-based food webs Ronald Benner Marine Science Institute, University of Texas at Austin, Port Aransas 78373

Jan Lay, Elizabeth K’nees, and Robert E. Hodson Department of Microbiology and Institute of Ecology, University of Georgia, Athens 30602 .4bstract Carbon conversion efficiencies were determined for the bacterial utilization of lignocellulosic detritus in waters from an estuarine and a freshwater wetland. Conversion efhciencies during bacterial growth on lignocellulose averaged -309’0 in both estuarine (salt marsh) and freshwater (Okefenokec Swamp) samples. Our estimates of bacterial growth efficiencies on refractory particulate detritus arc twofold to threefold higher than previous estimates owing, in large part, to the higher biovolumc-to-carbon conversion factor (0.22 g C cm-3) used in the present study to convert bacterial biovolumes into units of carbon. Bacterial growth cm lignocellulosic detritus was N limited in salt-marsh water and P limited in Okefenokec water; carbon conversion efficiencies increased to 450/0upon addition of ammonium and phosphate to salt-marsh and Okefenokee incubations, respectively. These results indicate that bacterial biomass produced at the expense of lignocellulosic detritus is likely to be an importan ( nutrient source to food webs in aquatic ecosystems with an abundance of macrophyte detritus and favorable conditions for microbial decomposition.

Particulate detritus derived from vascular plants is abundant in many freshwater and coastal marine ecosystems and has been considered an important source of nutrients to aquatic food webs (Odum and de la Cruz 1967; Mann 1972; Odum and Heald 1972; Brinson et al. 198 1). This detritus is compo-sed primarily of structural plant polymers, collectively referred to as li.gnocellulose (Hodson et al. 1984; Benner et al. 1985). Lignocellulose is indigestible by most aquatic animals and therefore is relatively unavailable for direct use by consumers in aquatic food webs. Our studies of the m icrobial transformations of vascular plant detritus indicate that primarily bacteria (relative to other microorganisms) are responsible for the degradation of lignocellulose in the aquatic environments we have examined (Benner et al. 1984b, 1986). Bacterial transformations of lignocellulosic detritus ——— Acknowledgments we thank W. J. Wicbe and two anonymous rCviCWers for helpful comments on an earlier version of this manuscript. P. A. Montagna assisted with statistical analyses, and R. T. Edwards supplied the fluorescent latex beads used to check the accuracy of our cell measurement technique. This research was supported by NSF Grant OCE 84-16384 and Grant NA 80 AADO0091 from the NOAA OffIce of Sea Grant.

may thereby function as an important link in the transfer of carbon and other nutrients from lignocellulose ‘to animals in aquatic food webs. In large part, the detritus-based nature of aquatic food webs in ecosystems that are dominated by vascular plants depends on the rates and efficiencies of conversion of lignocellulose to bacterial biomass. Rates of microbial transformation of lignocellulosic detritus are low relative to rates of transformation of the leachable, water-soluble components of vascular plant matter (Fallen and Pfaender 1976; Godshalk and Wetzcl 1978a,b; Benner and Hodson 1985), but the annual production of lignocellulose in many wetlands is high. Recent-studies of -the efficiencies of bacterial growth on particulate detritus from a variety of aquatic macrophytes indicate that bacterial growth efficiencies on these substrates are typically 10°/0or less (Stuart et al. 1981; Newell et al. 198 3a; Linley and Newell 1984). Such low carbon conversion efficiencies suggest that most of the detrital material is demineralized directly -and that only a small fraction of the particulate organic carbon from macrophytes flows further in food webs via microorganisms. In the present study we have further in-

1“514

Bacterial growth on detritus vestigated the role of bacteria in mediating the flow of nutrients from refractory macrophyte detritus to animals in aquatic food webs. Bacterial carbon conversion efficiencies on lignocellulose were investigated in waters from a salt marsh on Sapelo Island, Georgia, and from a sedge-dominated freshwater marsh in the Okefenokee Swamp, Georgia. The two marshes are similar in that they both are dominated by single species of herbaceous vascular plants, but rates of lignocellulose degradation by Okefenokee microflora are severalfold lower due to the low ambient PH (Benner et al. 1985). In addition to the large difference in ambient PH between these ecosystems, concentrations of inorganic nutrients also vary substantially. Therefore, the effects of ammonium and phosphate additions on lignocellulose carbon conversion efficiencies by bacterial assemblages from these wetlands were also investigated.

1515

lignin by weight; 77.70/0and 22.1 ‘/oof the radiolabcl in Spartina lignocellulose was associated with polysaccharides and lignin. Carex lignocellulose was 86.40/o polysaccharide and 13.6% lignin by weight; 78. 5?40 and 21. 50/0of the radiolabel in Carex lignocellulose was associated with polysaccharides and lignin.

Sample collection and incubation procedures – Water samples were collected from

March to November 1985 from a tidal creek on Sapelo Island and from the Okcfenokee Swamp. Water was used either as collected or was treated to remove or inhibit the activity of bacterivores. Bacteria were separated from eucaryotes by passing water samples through 1.O-~m pore-size Nuclepore filters (Benner et al. 1986). Cycloheximide (100 mg liter-l) was added to inhibit the activity of eucaryotes (Newell et al. 1983b; Furhman and McManus 1984; Sherr et al. 1986). Nutrient amendments were made to some water samples to investigate the reMaterials and methods lationship between inorganic N and P availability and bacterial carbon conversion efPreparation of radiolabeled lignocelluficiencies during growth on lignocellulose. lose – Short-form Spartina a[terniflora plants were collected from a salt marsh at Amended samples received either 140 ~gatoms liter-’ NH4+-N, 30 ~g-atoms liter-’ Sapelo Island. The sedge, Carex walteriana, or both. The concentrations of was collected from Mizell Prairie, a fresh- P043 ‘-P, water marsh in the Okefenokee Swamp. phosphate, ammonium, nitrate, and nitrite Plants were uniformly “C-labeled by ex- in water samples collected in June were deposing potted plants to an atmosphere con- termined with a Technicon autoanalyzer. Water samples (10 ml) were incubated taining 14COZ(Benner et al. 1984b). After labeling, the aboveground portions of the with [14C]lignocellulose (1 mg ml-’) in sterplants were removed and dried at 50°C. The ile glass bottles equipped with gassing ports. dried plant material was ground to pass a The microbial mineralization of lignocel40-mesh (425 ~m) screen and extracted with lulose was monitored by flushing the bottles organic solvents and water to remove the at 48-h intervals with sterile, humidified air nonstructural plant components, leaving an and trapping the evolved ‘4COZin a series extractant-free lignocellulose fraction (Benof two scintillation vials containing liquid ner et al. 1984a). This lignocellulose frac- scintillation medium (Benner et al. 1984a). tion consists primarily of the structural Radioactivity was determine&with a Beckpolymers (cellulose, hemicellulose, and lig- man LS 9000 liquid scintillation spectromnin) and constitutes 75-800/o (ash-free dry eter. The percentage of added radiolabel rewt) of the biomass of both Spartina and covered as 14COZwas used to calculate the Carex (Benner et al. 1985). total weight of lignocellulose-derived carThe radiolabeled plant material was bon that was microbially mineralized. Sparchemically fractionated to determine tina and Carex lignocelluloses were 40.72 whether the label was uniformly distributed and 45.71?40 carbon and 0.84 and 0.860/0nibetween the polysaccharide and lignin (acid- trogen by weight. Bottles were incubated at insoluble) components of the lignocellulose 25°C in the dark for 4 or 6 d. All incubations (Benner et al. 1984a). Spartina lignocelluconsisted of four or five replicates of each lose was 82.90/opolysaccharide and 17.1 Vo treatment.

1516

Benner et al.

Determination of bacterial abundance and biomass —Bacterial abundance was determined by epifluorescence microscopy with the acridine orange direct count (AODC) method (Hobbie et al. 1977). Samples that contained lignocellulose particles were scmicatedto remove bacteria from particles; sonication of samples for 90 s at a setting of 45 (Sonic Dismembrator model 300, Fisher Scientific) yielded the highest bacterial counts. We were concerned that all of the attached bacteria were not being accounted for by this method. Underestimation of bacterial numbers would artifactually decrease the calculated bacterial carbon conversion efficiencies. Therefore, our efficiency of counting bacteria in samples containing lignocellulose particles was estimated with two methods. One method used 3H-labeled bacteria (Hollibaugh et al. 1980) to determine the efficiency of removal of bacteria from lignocellulose particles (250-425 ~m). Natural bacterial populations (1-pm pore-size filtrate) from a salt-marsh creek and the Okefenokee Swamp were incubated for 2 d with fresh lignocellulose particles derived from Spartina and Carex, respectively. After 2 d [methyl-3H]thymidin-e (ICN Radiochemicals) was added (50 nM), and samples were incubated for 4 h to label the bacteria. Lignocellulose particles with attached 3H-labeled “bacteriawere sieved (26-~m pore-size screen) from the water and washed repeatedly with filter-sterilized water to remove unincorporated radiolabel. Free bacteria and unincorporated label were removed from the water by filtration through a 0.2-~m pore-size filter; the total amount of radiolabel incorporated into free bacteria was determined by liquid scintillation spcctrometry as described below. Particles with attached bacteria were resuspended in filtersterilized water and sonicated. After sonication the samples were sieved (26-~m pore size) to separate free bacteria from bacteria that remained attached to the particles. Free bacteria were filtered onto a 0.2-~m poresize Nuclepore filter and radioassayed; 1-ml portions of the filtrate were radioassayed to monitor the loss of radioactivity owing to cell rupture during sonication. The amount of radiolabel associated with free and at-

tached bacteria was determined by combusting filters and Iignocellulose particles to carbon dioxide and water (Benner et al. 1984a). The evolved 3H water was collected and radioassayed by liquid scintillation spectrometry. Radioactivity recovered from autoclave controls in each of the above fractions was subtracted from radioactivity recovered from live incubations. The amount.of radioactivity recovered after each treatment was totaled, and the percentage of the total radioactivity recovered in the free-bacteria fractions was considered to be the overall counting efficiency of bacteria. In the second method, natural bacterial populations (1-pm pore-size filtrate) from a salt-marshcreek and the Okefenokee Swamp were incubated for 2 d with fresh lignocellulose particles derived from Spartina and Carex, respectively. Lignocellulose particles and attached bacteria were sieved (26~m pore-size screen) from the water and thereby separated from free bacteria. Free bacteria w.erccounted by the AODC method. The particles were resuspended in filtersterilized water and sonicated a second time. After sonication the samples were sieved again (26-~m pore size) to separate free bacteria from bacteria that remained attached to the particles. Bacteria removed from particles during the second sonication were counted by the AODC method. Bacteria remaining on particles were also estimated by the AODC method, but background fluorescence made it difficult to accurately determine the number of bacteria remaining on particles. The number of bacteria recovered after each treatment was totaled, and the percentage of the total number of bacteria recovered in the free-bacteria fractions after the first sonication was considered to be the overall counting efficiency of bacteria. Results from .both methods indicated that the majority of the bacteria in the incubations, typically 50–70°/0, were unattached. The eiliciency of removal of attached bacteria from particles by sonication ranged from 20 to 50°/0. Of the two methods, estimating counting efficiency with radlolabeled ‘bacteriawas more conservative; overall counting efficiencies ranged from 65 to 70°/0. Overall counting efficiencies ranged

Bacterial growth on detritus from 80 to 900/0 in experiments with the AODC method. An average of the two counting efficiency estimates was used to correct all determinations of bacterial numbers in incubations with lignocellulose particles. This average counting efficiency of bacteria in incubations with Spartina lignocellulose was 80°/0, and in incubations with Carex lignocellulose it was 77Y0. The samples prepared for AODC were also used to determine bacterial volume (Fuhrman 1981). Cells were photographed on Kodak Ektachrome (400 ASA) 35-mm film. The sizes of w 300 bacteria from each sample set were measured by projecting the slides on a wall screen (5,700x magnification). Biovolumes were calculated from measurements of cell length and width and the formula: biovolume

= ~ W

()

L – $

where W is the measured cell width and L the measured cell length (L = W’ for cocci). Carbon content of bacteria was estimated with the conversion factor 0.22 g C cm-3 bacterial biovolume (Bratbak and Dundas 1984). In total, >16,000 determinations of bacterial biovolume were made. The accuracy of our estimates of cell dimension was checked with fluorescent latex beads with diameters similar to those of naturally occurring bacteria (Edwards 198 7). Beads were suspended in filter-sterilized water (= 10Gml-1, and filtered onto 0.2-~m pore-size Nuclepore filters prestained with Irgalan black. Filters with the fluorescent beads were viewed by epifluorescence microscopy, and bead diameters were determined as described above. The diameter of the beads, as stated by the manufacturer, was 0.73 ~m + 0.02 (*SD), and the calculated volume was 0.20 ~m3. The diameter of the beads, as measured by epifluorescence microscopy, was 0.67 ~m ~ 0.11, and the calculated volume was 0.17 ~m3. Carbon conversion e@ciency–All incubations received unamended or prefiltered natural water and therefore contained dissolved and particulate organic carbon that could serve as growth substrates for bacteria. Bacterial growth at the expense of these unlabeled pools of organic matter was de-

1517

termined in incubations that contained no added radiolabeled lignocellulose. Bacterial biomass produced in incubations with no added lignocellulose was subtracted from bacterial biomass produced in replicate incubations that received [ 14C]lignocellulose to determine growth eiliciencies on lignocellulose. Bacterial growth in salt-marsh water without added lignocellulose was minimal, whereas bacterial growth in Okefenokee water without added lignocellulose was substantial; subtracting the biomass produced in incubations without added lignocellulose was important in accurately determining the bacterial carbon conversion efficiencies at the expense of lignocellulose. Efficiency was corrected as follows: Carbon conversion efficiency

=

B .–BW (B,-,, – l?;+ (C=c as COJ

where BLCis bacterial biomass (pg C) produced in incubations with added lignocellulose, B ~ is bacterial biomass (~g C) produced in incubations without added lignocellulose, and CLC.iscarbon (pg C) from lignocellulose respired as COZ. In all treatments, the bacterial biomass produced in incubations with added lignocellulose was significantly greater (Student’s t-test, P < 0.05) than the bacterial biomass produced in parallel incubations without added lignoccllulosc.

Results Efects of inorganic nutrients on carbon conversion efficiencies —In the first series of experiments conducted in March 1985, prefiltered (1 -~m pore-size Nuclepore) water was incubated with [U- 14C]lignocellulose to determine the effects of inorganic nitrogen and phosphorus amendments on carbon conversion efficiencies of lignocellulose to bacterial biomass in the absence of the complicating effects of bacterivores. Adding ammonium (140 ~g-atoms N liter-1, to saltmarsh incubations had little effect on rates of Spartina lignocellulose mineralization to COZ, but did increase production of bacterial biomass from 10 to 15 ~g C d-1 (Fig. 1A). In contrast, adding ammonium to Okefenokee water had little effect on rates of either Carex lignocellulose mineralization

Benner et al.

r

SALT

MARSH

‘OKWij\OIEE n



4

— 1

8

3

6

2

4

1

2

0

0

40

40

30

30

11111 .— 2

Bacterial Biomass Respired IIS C02

20

20

10

10 ——

0

TREATMENT 1. Filtered water 2. Filtered water with ammonium

Fig. 1. Effects of added ammonium ( 140 pg-atoms N liter-1, on rates of lignocellulose degradation and on efficiencies of conversion of lignocellulosc to bacterial biomass in the absence of bactcrivores.

or bacterial production. Rates of lignocellulose mineralization in salt-marsh water were 10-fold higher than rates of lignocellulose mineralization in Okefenokee Swamp water. The higher rates of lignocellulose mineralization are due primarily to the higher pH (7) of salt-marsh water relative to the pH (4) of Okefcnokec water (Bcnner et al. 1985). Bacterial carbon conversion efficiencies on lignocellulose were similar (30.9 and a7.9°/0) in unamended salt-marsh and Okefenokee waters despite the large difference in overall rates oflignocellulose degradation between the two environments (Fig. lB). Adding ammonium to salt-marsh incubations increased bacterial carbon conversion efficiencies from 31 to 450/0but had no effect on conversion efficiencies of Okefenokee bacteria, indicating that bacterial growth efficiencies on lignocellulose were nitrogen limited in salt-marsh water but not in Okefenokee water. We did not measure the concentrations of ammonium in water samples on this date, but concentrations in estuarine waters off Sapelo Island typically range from 0.5 to 5 ~g-atoms N liter-1 (Imberger et al. 1983), and in Okefenokee water concentrations are typically in the range of 20-80 pgatoms N liter-’ (Moran et al. 1987).

1

.2

0 1

2

TREATMENT 1. Filtered water 2. Filtered water with ~hos~hate

Fig. 2. As Fig. 1, but of added phosphate (30 pgatoms P liter-’).

In a similar set of experiments conducted in April 1985, the effect of phosphate on bacterial carbon conversion efficiencies on lignocellulose was investigated. Adding phosphate (30 pg-atoms P liter-’) to saltmarsh incubations stimulated the rate oflignocellulose mineralization while production of bacterial biomass was unaffected (Fig. 2A). In contrast, phosphate additions to Okcfenokee incubations had little effect on rates of lignocell ulose mineralization but did increase bacterial production. As in :previous experiments, rates of lignocellulose mineralization were much lower in Okefenokee water relative to salt-marsh water. Bacterial carbon conversion efficiencies on ligrtocellulose in unamended, filtered saltmarsh water in April ( 17°/0) (Fig. 2A) were much lower than carbon conversion efficiencies determined in March (31 ‘/o). The addition of phosphate to salt-marsh incubations decreased the calculated carbon conversion efficiencies to 100/0 as rates of lignocellulose mineralization increased 1.5fold without a concomitant increase in bacterial production. Bacterial carbon conversion efficiencies in Okefenokee incubations, however, increased from 32 to 450/0in phosphate-amended samples, indicating that carbon conversion efficiencies on lignocellulose were phosphorus limited. Phosphate

1519

Bacteria[ growth on detritus

A

SALT MARSH

TREATMENT

TREATMENT

1. Unfiltered water

3. Unfiltered water with nutrients

1. Unfiltered water

3. Unfiltered water with nutrients

2. Filtered water

4. Filtered water with nutrients

2. Filtered water

4. Filtered water with nutrients

Fig. 3. Effects of bacterivores and of added ammonium and phosphate on rates of lignocellulosc degradation and on efficiencies of conversion of lignocelIU1OSC to bacterial biomass by natural populations of salt-marsh microorganisms.

concentrations typically range from 30 to 90 ~g-atoms P liter-’ in salt-marsh creek water (Pomeroy and Wiegert 1981) and from 5 to 15 Kg-atoms P liter-1 in Okefenokee water (Moran et al. 1987). Impact of grazing on calculated conversion efficiencies –The above-described experiments were conducted with natural water samples that had been size fractionated to remove bacterivores. The presence of active bacterivores during the incubation period will reduce the bacterial biomass present at the end of the incubation and thus artifactually decrease calculated carbon conversion efficiencies. A series of experiments was designed to estimate the impact of grazing on the calculated lignocellulose carbon conversion efficiencies of bacteria and to investigate the cfectiveness of size fractionation and antibiotic treatments in eliminating grazing activity. The accumulation of bacterial biomass was much higher in incubations with prefiltered (1-~m pore size) water than in incubations with unfiltered water, indicating that bacterial biomass produced at the ex-

Fig, 4. As Fig. 3, but on natural populations Okcfenokee Swamp microorganisms.

of

pense of lignocellulose was rapidly used by grazersin unfiltered samples (Fig. 3A). Adding inorganic nutrients to both filtered and unfiltered samples increased the production of bacterial biomass (Fig. 3A). Calculated growth efficiencies of bacteria on lignocellulose were much lower in incubations with grazers (3-1 4°/0)than in prefiltered incubations (36-45VO), demonstrating the importance of eliminating grazing in determ inations of growth efficiency (Fig. 3B). As demonstrated in the previous experiments, adding ammonium and phosphate increased bacterial carbon conversion efficiencies on Iignocellulose. The concentrations of nutrients in samples of Sapelo water collected in June were 5.5 pg-atoms liter-1 NH4+-N, 1.2 ~g-atoms liter-’ NO~---N + N02--N, and 35.9 pg-atoms liter-[ P043--P. Nutrient additions increased the concentration of ammonium 25-fold and the phosphate concentration 1.8-fold above ambient. A similar set of experiments was conducted with Okefenokec water samples. Rates of lignocellulose mineralization were much lower in filtered than in unfiltered samples (Fig. 4A). The effects of grazing on calculated carbon conversion efficiencies of

1520

Bennei- et al.

1

2

3

4

TREATMENT 1. Unfiltered water

3. Unfiltered water with cycloheximide

2. Filtered water

4. Filtered water with cycloheximide

Fig. 5. As Fig. 3, but of size fractionation and cycIoheximide additions on natural populations of saltmarsh microorganisms.

samples in addition to filtration to determine the relative effectiveness of the two methods in inhibiting grazers in salt-marsh water. In unfiltered water, rates of lignocellulose mineralization were similar in the presence and absence ofcycloheximide; rates of mineralization were lower in filtered samples but likewise were similar in the presence and absence ofcyclohexirnide (Fig. 5A). Both of the methods used to inhibit grazing —addition of cycloheximicle and filtration— resulted in greater accumulations of bacterial biomass by the end of the incubations (Fig. 5A). Bacterial carbon conversion efficiencies on Spiwtina lignocellulose in unfiltered saltmarsh water were much higher in August (22°/0;Fig. 5B) than in June (3°/0; Fig. 3B). Adding cycloheximide to unfiltered samples resulted j n an average calculated conversion efficiency of 280/o;the average conversion efficiency in filtered samples was 360/o(Fig. 5B). These results suggest that prefiltration of salt-marsh water through 1.O-pm poresize filters was more effective in reducing grazing pressure than was the addition of the antibiotic, cycloheximide. The addition of cycloheximide to filtered samples further increased the carbon conversion efficiency to 43(% (Fig. 5B), however, suggesting that filtration alone did not remove all grazers. Conversion efficiency estimates from all of these treatments are significantly different from one another at the 0.05 level (one-way ANOVA and Tukey’s multiple comparisons test).

Okefenokce bacteria were similar to those determined for salt-marsh bacteria. Calculated conversion efficiencies increased from 6°/0 in incubations with an active population of grazers to 240/oin incubations in which grazers were removed by prefiltration of water samples (Fig. 4B). Adding ammonium and phosphate increased rates of lignocellulose mineralization and bacterial conversion efficiencies in unfiltered water, but decreased conversion efficiencies in filSize-frequency distribution of bacteria — tered water. The concentrations of nutrients in samples of Okefenokee water coIlected in In unfiltered salt-marsh water that received June were 387.2 pg-atoms liter-i NHa+-N, Spartina Iignocellulose particles, the aver11.8 ~g-atoms liter-’ NO~--N + NOZ--N, age biovolume of cells after 6 d of incubation was 0.32 ~m3, and cells >0.4 pm3 ocand 11.4 pg-atoms liter-1 POq3--P. Nutrient additions increased the concentration of curred with the highest frequency (Fig. 6A). In ccmtrast, when grazers were removed by ammonium 1.4-fold and the phosphate prefdtration the average biovolume of cells concentration 2.6-fold above ambient. was 0.17 pm3, and cells with biovolumes in There have been several reports indicating that some grazers can pass through filters the range of 0.01-0.10 ~ms were the most prevalent and accounted for 4“8°/o of the total with pore sizes as small as O.6 pm (Fuhrman and McManus 1984; Cynar et al. 1985). We bacterial population (Fig. 6B). Differences between the size-frequency distributions of considered that we could be underestimating bacterial carbon conversion efficiencies bacteria could result from selective grazing if grazers remained in our filtered samples. pressure on smaller cells in unfiltered water To test this possibility we ran a set of ex- or from prefiltration of water samples, as a significant number of larger bacteria preperiments in August in which the eucaryotic inhibitor, cycloheximide, was added to some sumably are retained on 1.O-prn pore-size

1~

1521

Bacterial growth on detritus IA n

I

I

Unfiltered water with Iignocetlulose

particles

30

,:.,.,:,:, :::.,:: ....... ..,,.,,,:,: W :: ~::: .,.,.,, :?.;,:: :.

~\o 25 w

z 20 s? 1a

m

6

IEI

:;:: :::::: ;:::::::: ?.::::::: I OKEFENOKEE SWAMP

%

15

IEl

Unfiltered water with

Iignocellulose

particles

E%NM’llle

10

~5 1mo

z

Filtered wakv will] lignocellulose

:::;: ;;:,: :.:,:.:,:.: ,.,:,:,:,:, ,>.,:,:,,.,

particles

c

30

Ml :,. :

z W 25

Filtered water with lignocelluloso

particles

3 020

w ~

15 10 5 ()

12345678910

BIOVOLUME

1

(pm3)

Fig. 6. Size-frequency distributions of bacteria in salt-marsh water with added Spartina lignocellulose particles.

filters. The average biovolume of bacteria in incubations with unfiltered water that received cycloheximide to inhibit grazers was 0.25 pm3, suggesting that selective grazing of smaller cells did contribute to the greater average biovolume in unfiltered samples. The average biovolume of bacteria was always greater in the presence of lignocellulose particles than in the absence of added particles, and rod-shaped bacteria predominated. A somewhat different size-frequency distribution pattern was observed in Okefenokee samples. In unfiltered Okefenokee water with added Carex lignocellulose particles, the average biovolume of cells after 6 d of incubation was 0.17 pm3, and small cells in the range of 0.01-0.05 pm3 occurred with the highest frequency (Fig. 7A). Prefiltration of water samples and incubation for 6 d with lignoccllulose particles resulted in an increase in the number of cells in the size range from 0.05 to 0.15 ~m3 and a corresponding decrease in the smallest and

2345678910

BIOVOLUME

(pm3)

Fig. 7. As Fig. 6, but in Okefenokee Swamp water with added Carex lignocellulosc particles.

largest size ranges (Fig. 7B). The average biovolume of the bacterial population, however, did not change. As with salt-marsh samples, the average biovolume of Okefenokee bacteria was always greater in the presence of lignocellulose particles than in the absence of added particles, and rodshaped bacteria predominated.

Spec&

growth rates on ZignoceZ[uZose–

Values for specific growth rates (Tables 1 and 2) were based on changes in bacterial biomass rather than changes in bacterial numbers. These values are representative of the total bacterial growth in the incubations and growth at the expense of the added lignocellulose. Most (87-990/o) of the bacterial growth that occurred in incubations with salt-marsh water resulted from utilization of the added Iignocellulose. Bacterial growth on naturally occurring dissolved organic matter (DOM) in salt-marsh water was relatively minor in these incubations, and the specific growth rates for the total bacterial growth and growth on lignocellulose were similar (Table 1). Specific growth rates on

1522

Benner et al.

Table 1. Specific growth rates (v) and growth effiTabIe 2. As Table 1, but for Okefenokec bacteria cienciesforsalt-marshbacteria(l -~mpore-sizefiltrate) incubated with Carex walteriana Iignocellulose. keubated witfs Spartina a{terrufiora fignocellulose Carbon ‘(N.D.–not determined). Growth on Total growth 1985*

~(h

‘)

lignocdlulosc

Carbon Total Iignocel- conversion IUIOSC-C used

P (h ‘)

(I’@

efficiency (%)

Mar Apr Jun

N,D, 0.0235 0.0300

0.0264 0.0218 0.0178

125.6 179.0 149.5

30.9 17.3 36.5

Aug —— * March

0.0245

0.0237

182.9

35.9

results are from 4-d incubations, for o{her months results are from 6-d incubations.

Total growth 1985*

p(h))

Mar Apr

N.D. 0.0298

Jun Nov

0.0303 0.0251

Growth on lignocellulosc w (h ‘)

Total lignocel- conversion lUIOSC-Cused etliciency k)

(v.)

0.0078

12.3

27.9

0.0073 0.0030

34.7 19.1

32.1 24.1

0.0081

12.5

32.3

* March

results are from 4-d incubations, for other months results arc from 6-d incubations.

ciencics (9- 14°/0)of Spartina detritus (FJewell et i~l. 1983a). One factor is responsible for a large part of this discrepancy in estiSpartina lignocellulose ranged from 0.0178 mates of bacterial growth efficiencies on reto 0.0264 h-’ insalt-marsh incubations, and carbon conversion efficiencies ranged from fractory particulate detritus. Microscopic 17 to 370/0at natural nutrient concentrations determinations of bacterial numbers and biovolumes were used in both studies to (Table 1). The average carbon conversion estimate bacterial growth, but values used efficiency of lignocellulose was 300/0in saltto convert bacterial biovolume to bacterial marsh water. carbon content varied by a factor of two. Bacterial growth at the -expense of naturally occurring DOM was much more sign- Newell et al. (1983a) used a factor of 0.11 g C cm-3. The validity of this conversion ificant in Okefenokee water than in saltmarsh water. During the spring and fall factor, which was calculated from values of the buoyant density, the ratio of dry weight sampling periods most of the bacterial to wet weight, and the ratio of carbon weight growth (66–7 1‘/o) in Okefenokee incubations resulted from use of the added lig- to dry weight of bacteria].cells, was recently challenged by Bratbak and Dundas (1984). nocellulose, but during the summer samThey demonstrated that previous converpling period only 360/o of the measured bacterial growth in. Okefenokee incubations sion factors were in error because cor-rec. tions for intercellular water were not made resulted from use of the added l.ignocelluIose,, The lower specific growth rate on lig- in determining the ratio of dry weight to wet nocellulosc during summer (Table 2) is at- weight o:f bacterial cells and advocated a conversion factor of 0.22 .g C cm-3, as we tributed to a greater abundance of readily useci DOM. The concentrations of DOM. in used in the present study. More recent studies using the same methOkefenokee waters are quite high (30-50 mg C liter-’; Flebbe 1982) and are known to ods as our-sto determine bacterial volumes support moderately high rates of bacterial have estimated even higher conversion facproduction (Murray and Hodson 1985). tors, ranging from 0.38 to 0.43 g C cm-3 Specific growth rates on Carex lignocellu- (Bratbak 1985; Lee and Fuhrman 1987). lose ranged from 0.0030 to 0.0081 h-~ in Bjm-nscn ( 1986) used an image analysis sysOkefenokee incubations, and carbon con- tem to estimate bacterial biovolumes anti version efficiencies ranged from 24 to 32% determined a conversion factor of 0.35 g C at natural nutrient concentrations (Table 2). cm-~; Kogurc and Koike (1987) used a par‘The average carbon conversion efficiency of ticle counter to size bacteria and determined lignocelhdose was 29% in Okefermkee water. a conversion factor of 0.2 g C cm-3. It is now apparent that current estimates of bacDiscussion terial volume-to-carbon conversion factors are twofold -to threefold higher than previThe carbon conversion efficiencies (17y accepted values and that carbon con37°!o)determined in this study for the bac- OUSI terial utilization of Spartina lignocellulose version efficiencies calculated with earlier are substantially higher than previous esti- conversion factors significantly underestimates of bacterial carbon conversion efli- mated bacterial growth yields on refractory

Bacterial growth on detritus particulate detritus. We chose a conservative conversion factor in this study as the bacterial cells in our incubations were fairly large, and several investigators have noted that larger cells tend to have less carbon per unit biovolume than smaller cells (Lee and Fuhrman 1987; Norland et al. 1987). Correction of the carbon conversion efficiencies estimated by Newell et al. (1983a) to account for the higher carbon content of bacteria per unit volume yields estimates of 1828?h–values that are in general agreement with ours (17-370/0). Similar estimates of carbon” conversion efficiency (22–430/o) for bacterial growth on seaweed detritus were made with direct measurements (CHN analysis) of bacterial carbon rather than values derived from biovolume determinations (Robinson et al. 1982). The removal or inhibition of grazers of bacteria during incubation of water samples with plant detritus is critical in accurately determining carbon conversion efficiencies. In the presence of active populations of grazers, bacterial carbon conversion efficiencies on lignocellulose were underestimated by as much as a factor of 12. Removal of grazers from water samples by filtration was more effective in reducing grazing pressure than was the addition of cycloheximide. It appeared, however, that filtration alone did not completely remove grazers, and therefore the carbon conversion efficiencies calculated from prefiltered samples may be underestimates of actual conversion efficiencies. Carbon conversion efficiencies of 430/owere calculated from saltmarsh samples that were both prefiltered and treated with cycloheximide, relative to conversion efficiencies of 36V0in samples that were only prefiltered. Prefiltering water samples or adding cycloheximide could alter microbial growth rates but are unlikely to affect our estimated conversion efficiencies, since bacterial biomass produced in parallel incubations without added lignocellulose was subtracted from the biomass produced in incubations that received lignocellulose. A somewhat different approach to estimating bacterial carbon conversion efficiencies on phytoplankton and macrophyte detrituswas used by Newell et al. (1981, 1983a) and Stuart et al. (1981). They monitored

1523

protozoan as well as bacterial populations in unfiltered water samples and made estimates of conversion efficiencies during periods of logarithmic bacterial growth when grazer populations were low. Newell et al. (1983a) estimated that