SINSABAUGH, ROBERT L., STUART FINDLAY, PAULA FRANCHINI

Transcription

SINSABAUGH, ROBERT L., STUART FINDLAY, PAULA FRANCHINI
Limnol.
0
1997,
Ocecmogr., 42(l),
by the American
1997, 29-38
Society
of Limnology
and Oceanography,
Inc.
Enzymatic analysis of riverine bacterioplankton production
Robert L. Sinsabaugh
Biology Department, University
of Toledo, Toledo, Ohio 43606
Stuart Findlay
Institute of Ecosystem Studies, Box AB, Millbrook,
New York 12545
Paula Franchini
Biology Department, University
of Toledo
David Fischer
Institute of Ecosystem Studies
Abstract
Bacterial production is the entry point for detrital macronutrients into aquatic food webs. Many factors affect
productivity, but the heterogeneity of detrital substrates and the diversity of microbial communities confound simple
relationships between carbon supply and growth. WC tried to link the two by analyzing extracellular enzyme activities. Water samples were collected from three rivers and assayed for bacterial productivity and the activities of
eight enzymes. Production varied among systems, peaking at 644, 170, and 68 pmol C liter-’ d-’ in the Ottawa
(Ohio), Maumee (Ohio), and Hudson (New York) Rivers. V,,,,,values were generally correlated with productivity.
The mean ratios of productivity per unit peptidase and esterase activity were similar among rivers, whereas carbohydrase and phosphatase ratios varied widely. The data were used to evaluate a model that relates productivity
to carbon flow by using enzyme activities as indicators and assuming an optimum resource allocation relationship
among C-, N-, and P-acquiring enzymes. The data supported the model, but predictive power was low. Bacterial
productivity generally increased with inorganic nutrient availability, but high levels of productivity at any specific
eutrophic state required sources of both saccharides and amino acids.
been applied to DOC (Thurman 1986; Hedges et al. 1994;
Moran and Hodson 1994) to trace its origin, composition,
and fate, but connections to bacterial productivity have been
difficult to demonstrate. Direct studies of bacterial productivity in relation to DOC have involved productivity assays
in conjunction with analyses for specific labile DOC components, such as amino acids and sugars. At its highest resolution, this approach focuses on the kinetics of monomer
generation and uptake, rather than on simple abundances.
Both of these processes are mediated by ectoenzymes. The
activity of enzymes involved in monomer generation is typically analyzed using artificial substrates linked to chromogenic or fluorogenic moieties (Hollibaugh and Azam 1983;
Hoppe 1983; Someville and Billen 1983). The activities of
membrane permeases, which collectively constitute assimilation, are usually studied with radiolabeled monomers
(Wright and Hobbie 1966). Such research has shown that
monomer generation is generally slower than assimilation;
therefore, the rate-limiting step in bacterial macronutrient
procurement is usually the generation of assimilable monomers from polymeric or condensed DOC components
(Chrost 199 1).
There is a gap between high-resolution biochemical models and larger scale spatiotemporal patterns of bacterial productivity (e.g. Kaplan and Bott 1983, 1985; Bott et al. 1984).
If ectoenzymatic breakdown of complex organic molecules
is the rate-limiting process in bacterial carbon acquisition, it
may be possible to analyze productivity patterns in relation
In many aquatic ecosystems, a large fraction of macronutrient flow is through the microbial loop, a cycle that recovers detrital carbon and other nutrients and injects them
back into the food web (Pomeroy 1974; Azam et al. 1983).
Dissolved organic carbon (DOC) is assimilated into the microbial loop through heterotrophic bacterial production.
Among ecosystems, bacterial production can be correlated
with DOC concentration or, more indirectly, with primary
production (Cole et al. 1988). Within an ecosystem the trophic basis of bacterial production is more difficult to discern
(Findlay et al. 1991), partly because other physicochemical
[UV radiation (Herndl et al. 1993), nutrients (Meyer-Reil
1987), temperature (Hudson et al. 1992; Shiah and Ducklow
1994)] and biological factors [grazing (Gonzalez et al. 1990;
Bott and Kaplan 1990), growth efficiency (Kroer 1993; Biddanda et al. 1994)] play a role in regulating productivity.
Also, because DOC is a heterogeneous mix of compounds,
gross abundance does not equate with biodegradability. Additionally, DOC and production dynamics may be out of
phase due to lags in bacterial response to changes in the
DOC milieu (Scavia and Laird 1987).
Various Chromatographic and spectrometric analyses have
Acknowledgments
This work was supported by grants from the Hudson River Foundation and the U.S. Dept. of Agriculture. Chemical analyses of Ottawa River water were conducted by Sarah Karpanty, Sara Mierzwiak, and Mark Robinson. Christine Foreman and William Sobczak assisted with the enzyme and productivity assays.
29
30
Sinsahaugh et al.
Table I. Water quality measures for the Ottawa River. Data were
collected at approximately biweekly intervals from April to December 1994. (ND-not detected.)
Parameter
Mean (SD)
Temp (“C)
PH
Alkalinity (mg liter ‘)
NH,+ (mg liter-‘)
NO,- (mg liter-l)
Cl- (mg liter ‘)
S0,2- (mg liter-‘)
Soluble reactive PO,?(pg liter-‘)
DOC (mg liter-‘)
15
7.7(0.3)
197(20)
0.2(0.2)
7.7( 12.2)
138(50)
123(40)
2-26
7. I-8.1
176-229
ND-O.7
ND-39.6”
40-225 t
65-200-j.
31
20
5
20
20
18
18
107(65)
6.9(2.1)
ND-2 10
4.7-l 0.3
5
29
Range
N
* Nitrate was high in spring owing to agricultural runoff, then remained low
through summer and autumn.
-1Chloride and sulfate concentrations increased markedly in late autumn,
coincident with salt application to streets and parking areas.
to these activities to gain a broader understanding of this
carbon transduction process. To evaluate this approach, we
followed the seasonal dynamics of bacterioplankton in three
rivers.
Methods
Site descriptions-The
Ottawa River drains a 446-km2,
low-gradient, agricultural catchment in northwestern Ohio,
discharging into Lake Erie near Toledo. Our sampling site
was located on the University of Toledo campus, -8 km
from the mouth. At this point, the channel is - 10 m wide
with a soft sediment substratum. A riparian strip, which extends several kilometers upstream, shades the channel during
the growing season. At base flow, the current is negligible
and water depth averages 30 cm. Because of the low gradient, a large fraction of annual discharge and transport is
associated with runoff events. Water quality data, measured
at the sampling site, are summarized in Table 1.
The Maumee River is a large, low-gradient stream (mean
annual discharge of -140 rn” s-l) that enters Lake Erie at
Toledo. Most (76%) of its 17,000-km2 catchment is used for
corn and soybean production (Baker 1993). Water samples
were collected from two stations: International Park in
downtown Toledo, located near the mouth of the river; and
the raw water intake for the town of Bowling Green, -20
km upstream. Water quality at the latter station has been
monitored for ~15 yr (Baker 1993). The flow-weighted
mean concentration of suspended solids is 211 mg liter-l; of
total P, 0.44 mg liter-‘; soluble reactive P, 0.07 mg liter-‘;
nitrate plus nitrite, 5.4 mg liter-‘; total Kjeldahl N, I .9 mg
liter-‘; chloride, 28.7 mg liter-‘; and DOC, 7-8 mg liter-’
(October 1994-April 1995).
Samples from the Hudson River were collected from a
well-studied
station
near Kingston,
New York
(Cole et al.
1992; Findlay et al. 1991). This location is tidal (range, l2 m) but well above any influence
of salt water. Inorganic
nutrient concentrations are moderate to high, with dissolved
inorganic N -70 PM and P -1 PM.
Sampling-Ottawa
River water samples were collected at
I-2-week intervals from March through December I994 by
lowering a bucket from a foot bridge at midchannel. Because
the sampling site was adjacent to the laboratory, analyses
commenced immediately after collection. Water temperature
was recorded and subsamples were removed for water quality measurements. Measurements of absorbance at 254 nm,
a surrogate for DOC concentration, were made on subsamples passed through Gelman A/E glass-fiber filters that were
prerinsecl with deionized water. These values were later converted tc DOC concentrations (mg liter-l) by using a conversion fjctor of 40, determined empirically by sending selected samples to the Institute of Ecosystem Studies for DOC
analysis by means of high-temperature catalytic oxidation
(Shimadzu 5000). Water chemistry measurements, except
phosphorus, were made with Hach Test Kits. Soluble reactive phosphorus was determined by means of the ascorbic
acid assay (Am. Public Health Assoc. 1992). pH was measured electrometrically and alkalinity by titration to pH 4.2.
The Maumee River stations were sampled approximately
biweekly from April through November 1995. At the downstream station, water was collected from a pier by bucket.
At the upstream station, water was collected directly from
the water intake line. These water samples were returned to
the labor,atory for analysis within an hour of collection.
Hudso.1 River water samples were collected biweekly
(April-D’zcember
1994) adjacent to the main navigation
channel in conjunction with routine sampling. Water was
collected from 1 m beneath the surface with a peristaltic
pump. The entire water column (depth, -6 m) is well mixed
with no vertical gradients of temperature or oxygen. Water
samples for enzyme assays were refrigerated until analyzed,
usually the day after sampling.
Enzyrnc: assays-to
examine macronutrient acquisition by
planktonic microbiota, we assayed water samples for the activity of several ectoenzymes: fatty acid esterase (FAEase; EC
3.1.1.1), leucine aminopeptidase (LAPase; EC 3.4.11.1),
“trypsin”
endopeptidase (TEPase; EC 3.4.21.4), alkaline
phosphatase (APase; EC 3.1.3.1), a- 1,4-glucosidase (cwGase;
EC 3.2.1.20), p- 1,4-glucosidase @Gase; EC 3.2. I .21), p-xylosidase (PXase; EC 3.2.2.37), and P-ZV-acetylglucosaminidase (NAGase; EC 3.2.1.30). The substrates for these
(MUF)
enzymes were, respectively, 4-methylumbelliferyl
acetate, L-leucine 7-amido-4-methyl-coumarin, MUF-p-guanidinobenzoate, MUF-phosphate, MUF-a-D-glucoside, MUFP-D-glucoside, MUF-P-D-xyloside, and MUF-N-acetyl-P-glucosaminicle. Substrate solutions were prepared in 5 mM
bicarbonate buffer (pH 8).
At the University of Toledo, the assays were performed
in black 96-well microtiter plates at ambient room temperature (-21.1°C). Each well received 200 ~1 of unamended
river water and 50 ~1 of sterile substrate solution. Fluorescence realdings from each microplate were recorded at 0. I l-h interirals, depending on the amount of activity present,
with a Dynatech MicroFLUOR platereader with a 365-nm
broadband excitation filter and a 450-nm narrowband emission filter Activities were expressed as the rate of accumulation of methylumbelliferone equivalents (pm01 liter-l h -I)
by using an emission coefficient of 4,100 nmol-I (calculated
Bacteria production in rivers
by regression from 0.25-3.5 PM MUF standards in 5 mM
bicarbonate). Individual samples were corrected for quenching, which ranged from <lo to 40%.
Enzyme kinetics were assumed to follow the MichaelisMenton model. The kinetic parameters K,, (the half-saturation constant) and V,,,,, (maximal velocity of reaction, measured at substrate saturation) were estimated from linear regressions of the Eadie-Hofstee equation, a linearized form
of the Michaelis-Menton model:
v = -K,,(V/[L!Tj) + l/V,,,,.
V is the reaction rate (nmol h-l ml-l) and [S] is the substrate
concentration (nmol ml -I). For each enzyme, V was measured at eight substrate concentrations with three analytical
replicates per assay. The range of substrate concentrations
used varied among enzymes and over time: cxGase and
PGase, 2-40 PM; LAPase and TEPase, 5-l 20 PM; APase,
l-1 20 PM; FAEase, l-40 PM. The goal was to select a
range of substrate concentrations that encompassed K,,,; however, during periods of high microbial activity, the apparent
K,,, of LAP and FAE sometimes exceeded substrate solubility.
At the Institute of Ecosystem Studies (IES), enzyme assays were conducted in white 96-well microtiter plates by
using a Perkin-Elmer fluorometer. Other assay conditions
were the same as those used at the University of Toledo,
except that, the nanomolar emission coefficient for methylumbelliferone was 589. Kinetic analyses were not done; the
enzyme activity dataset includes only V,,, values, obtained
by measuring V under substrate saturating conditions.
Labile substrate pools-For
the 1994 Ottawa River samples, kinetic analyses of the ectoenzymatic hydrolysis of artificial, MUF-linked substrates were used to make indirect
estimates of the turnover time and, in some cases, the size
of the resident labile substrate pools (LSPs) associated with
individual enzymes. LSP turnover time, TO, was calculated
from linear regressions of MUF substrate turnover (T, in
hours) vs. MUF substrate concentration ([S] in nmol ml-‘):
T = TO + [Silk,,,
where k,sis a rate constant (nmol h-l ml I).
For APase and PGase, we estimated LSP concentration
by running kinetic assays on Ottawa water samples that had
been spiked with known concentrations of glucose-6-phosphate (10 pm) or cellobiose (20 pm). The increment in LSP
turnover time produced by the addition of a model competitive inhibitor was used to calculate LSP concentration:
WPI = KlWG7 - To).
[a is the final concentration of the model inhibitor, T,, is the
turnover time of the LSP and T, is the turnover time of the
LSP in the presence of the model inhibitor. Half-saturation
constants, K,, for the model inhibitors were calculated from
shifts in the apparent K,,, of the MUF substrate reactions:
K, = VlWwz’K,, - 1).
Km’ is the apparent K,,, for the MUF substrate reaction in the
river water amended with model inhibitor.
The [LSP] estimates are contingent on the assumptions
31
that the competitive inhibitors accurately model the constituents of the resident substrate pool and that the addition of
competitive inhibitors to the resident substrate pool did not
significantly increase LSP reaction rates.
Bacterial productivity-Bacterioplankton
secondary productivity was estimated from the rate of incorporation of
[“HIthymidine (TdR) into DNA: lo-ml samples were placed
in 15-ml polypropylene tubes with 1.5 MBq [methyl3H]thymidine (sp act, 1.2 TBq mmol-‘; final thymidine concentration, 57 nM), vortexed, and incubated for 1 h at 20°C.
There were four analytical replicates and two negative controls for each sample. Negative controls were water samples
that had been boiled for 30 min or to which Formalin had
been added prior to TdR addition. After incubation, cells
were collected by vacuum filtration on 47-mm-diameter, 0.2pm membrane filters, washed twice with lo-ml aliquots of
cold 5% TCA, and frozen for later DNA extraction.
DNA was extracted by digesting each filter for 15 min at
25°C in 5 ml of extraction solution (12 g liter-’ NaOH, 7.5
g liter-’ EDTA, 1.0 g liter-’ SDS). Following digestion, the
tubes were centrifuged and the supernatants transferred to
clean tubes in an ice bath. A 0.5-ml aliquot of 3.0 N HCl
was added to neutralize NaOH, followed by 0.7 ml of 50%
TCA. After 15 min, 0.1 ml of herring sperm DNA (10 mg
ml-l) was added to each tube to promote precipitation.
Precipitates were pelleted by centrifugation at 20,000 X
g for 15 min at 0°C. The supernatants were discarded and
the pellets were resuspended in 3 ml of 5% TCA and centrifuged again. After a second 5% TCA wash, the precipitated DNA was hydrolyzed by mixing the pellet with 3 ml
of 5% TCA and placing the tubes in a dry bath at 95°C for
30 min. The digestion mixture was centrifuged at 5,000 X
g for 5 min; 1.O-ml aliquots of supernatant were transferred
to scintillation vials containing 10 ml of counting cocktail.
Each vial was counted for 10 min or 2% error. Counting
efficiencies were determined by external standardization. Results were initially calculated as bacterial cells produced h-l
ml-’ by means of a conversion factor of 2 X 10” cells produced nmol-’ of TdR incorporated (Riemann et al. 1990)
and then converted to carbon units by means of a conversion
factor of 28 fg C cell-’ (Findlay et al. 1992). Isotopic dilution trials were conducted at both sites. Dilution effects were
not statistically significant for Ottawa and Maumee River
samples; however, they were for Hudson River samples
(Findlay et al. 1991), and productivity values were adjusted
accordingly.
MARCIE model--The enzyme and productivity data were
used to evaluate the MARCIE (Microbial Allocation of Resources among Community Indicator Enzymes) model for
bacterioplankton. This model was originally developed to
describe the enzymatic activity of microbial communities in
relation to plant litter decomposition (Sinsabaugh and Moorhead 1994). It is based on the premises that ectoenzymatic
hydrolysis of complex molecules is the rate-limiting step in
bacterial production; that ectoenzyme synthesis is controlled
at the level of transcription by induction and repressionderepression mechanisms that are linked to environmental
nutrient availabilities (Chrost 1991); and that at the com-
32
Sinsabaugh et al.
munity level, these regulatory processes manifest as an optimal resource allocation strategy.
In the MARCIE model, ectoenzymes are grouped into
three categories-those
principally involved in carbon acquisition (Ec), those involved in nitrogen acquisition (EN),
and those involved in phosphorus acquisition (E,). Because
of their common regulatory motif, it is assumed that the
activities of the many enzymes in each group are sufficiently
correlated such that the activity of one, or preferably a few,
enzymes in each group can act as an indicators for the whole
group. For bacterioplankton the basic form of the model is
P = kcEcz,
where P is production rate and kc is a first-order rate constant
(production per unit E, activity). The model predicts that
production is proportional to C flow, whose availability is a
function of EC activity. However, the production of C-acquiring enzymes is constrained by the need to enzymatically
acquire N and P; thus, EC can also be expressed as a fraction
of total extracellular enzyme production (E,, the sum of EC,
EN, and E,,) whose value is dependent on N and P availability
(see Sinsabaugh and Moorhead 1994 for complete derivation):
P = k,E,l( 1 + EN/EC + EJE,)
Evaluating the model involves testing two hypothesesthat production rates are directly proportional to EC activity,
and that C, N, and P acquisition activities are linked through
an optimum resource allocation strategy (specifically that P
is directly related to EC/EN and EC/E,). The null model tests
whether P is directly related to E,, under the hypothesis that
production reflects total extracellular enzyme activity and
that ET can increase or decrease without reallocating resources among EC, EN, and Ep (e.g. through changes in
growth efficiencies).
To evaluate the model, we transformed the activity of each
enzyme to a O-l scale by dividing each value by the highest
value in the dataset. This standardization eliminates scaler
weighting among enzymes. Each enzyme is assigned to one
of the three enzyme groups. If there is more than one enzyme
in a group, the activities of the set are averaged. Thus, the
maximum value for EC, EN, and Ep is 1 and ET cannot exceed
3. The rate constant kc is the slope of the regression P vs.
ET/( 1 + EN/EC + EJEc). The hypothesis of EC control is
accepted if this regression has an F statistic that is significantly greater than that of the null model P vs. ET. The
resource allocation hypotheses are accepted if the regressions P vs. EJE, and P vs. EC/E, are statistically significant.
Results
Enzyme kinetics are typically described by the MichaelisMenton model for the reaction between a simple enzyme
(i.e. nonallosteric) and its substrate. This model does not
always describe the collective behavior of heterogeneous enzyme populations distributed among a diverse community of
organisms. We used the linear Eadie-Hofstee transformation
of the Michaelis-Menton equation to estimate V,,,, and K,,,
parameters. In most cases, the data fit the model well, but
some enzymes were more problematic than others. FAEase
had the lowest goodness-of-fit statistics (mean ti = 0.55).
These weak fits were at least partially due to the instability
of the MUF-acetate substrate, which decomposed even in
sterile buffer, leading us to doubt its specificity as a measure
of lipase activity. At the other extreme, LAPase gave the
best fits to the Eadie-Hofstee model (mean r2 = 0.81).
In all three of the rivers monitored, enzyme activities and
associated kinetic parameters showed considerable fluctuation (Table 2). V,,,, values for crGase, PGase, and APase
were the most variable for Ottawa and Maumee River samples. In the Hudson River, /3Xase was the most variable,
followed closely by aGase and APase. In all systems, FAEase activity displayed the least variability (range <lo). The
highest I”,,,, values were measured in the esterase and peptidase assays, the lowest in the carbohydrase assays (Table
2). In general, apparent K,, values were more stable than V,,,
and To parameters. Among assays, the two Michaelis-Menton parameters tended to scale together. The highest Km values, like V,,,,,, were associated with the esterase and peptidase assa.ys;values for the carbohydrases were generally one
to two orders of magnitude lower. Within enzymes, however,
Kl and Klax were significantly correlated only for FAEase,
TEPase, and LAPase (Table 3). The differences in K,, with
V,,, correlations among enzymes may reflect differences in
lag time, i.e., the interval between changes in the size of the
labile substrate pool and enzyme response achieved through
the production or turnover of enzymes.
The constant Km can be used as an indicator of the relative
size of a LSP for two reasons. First, the LSP competes with
added anificial substrate for the active sites of the enzyme,
so increases in [LSP] manifest as increases in apparent Km.
However, an increase in measured K,,, may also reflect an
increase in the true K,, of the enzyme population, i.e. the
substrate concentration at which half the catalytic sites are
occupied. In this case too, a high K,, suggests that the [LSP]
is large relative to bacterioplankton requirements; otherwise,
competition for limiting nutrients should select for lower K,,
values. T’hus, the high correlation between apparent K,, and
V,,, for fatty acid esterase (r = 0.85, Table 2) indicates a
close relationship between the size of the labile substrate
pool and the capacity of the planktonic community to utilize
the pool. Comparisons of K,, and V,,,,, over time for TEPase
and LAP#ase (r = 0.39 and 0.33, respectively) suggested that
there may be a temporal lag between changes in [LSP] and
the response of the community (V,,,,) to those changes. However, our data are not of sufficiently high resolution to estimate these lag times.
For APase and PGase, we were able to estimate the halfsaturation constant (K,) and labile substrate pool size using
glucose-6-P and cellobiose, respectively, as model “natural”
substrates. In general, the results corroborated those from the
MUF substrates. For APase, K, for glucose-6-P ranged from
5.1 to 40.5 PM (mean = 18.0, n = 18) and [LSP] (in glucose-6-P equivalents) ranged from 2 to 1,015 ,zM (mean =
118, n = 16); [LSP] was significantly correlated with apparent K,, (r = 0.83). For PGase, KI for cellobiose ranged
from 1 to 46 PM (mean = 1 I, n = 18) and [LSP] (in cellobiose equivalents) ranged from 1 to 322 ,uM (mean = 52,
n = 16); Kl was significantly correlated with apparent K,, (r
33
Bacteria production in rivers
Table 2. Summary statistics for enzyme activity parameters in the Ottawa, Maumee, and Hudson
Rivers. The units for V,,,, are pmol d I liter-‘, for K,,, are pm01 liter-‘, and T,, are h (n = 24 for
Ottawa River, 30 for Maumee River, 18 for Hudson River). (ND-not detected.)
Enzyme
Parameter
FAEase
LAPase
Vmnx
K
TEPase
APase
cvGase
PGasc
VIllaX
K,,
T”
Ottawa
Maumee
0.6-10.7(3.9)
4-l 15(36)
3-396(28)
0.13-3.4(1.3)
22-15 1(67)
19-780( 124)
0.5-10.3(3.0)
18- 140(54)
12-48(25)
0.09-6.5(0.7)
1.5-10.1(4.1)
2-182(70)
0.0 l-2.4(0.3)
0.12-2.2(0.8)
l-535(74)
0.01 l-1.37(0.2)
O.l-14.6(2.2)
0.3-918(127)
l-3-6.6(2.9)
6.6-32.0( 16.9)
3.7-20.4(9.0)
0.2-2.2(0.7)
10.6-104.5(41.8)
15-495(94)
1.8-21.6(5.6)
7.9-268.8(7.9)
8-503(53)
0.01-l .03(0.2)
1.3-50.2(6.3)
2-3,840(234)
0.002-0.04(0.02)
0.16-6.6( 1.9)
3-12,800(1,100)
0.001-0.1 l(O.02)
0.005-10.3(2.4)
30-3,460(550)
= 0.61). The mean [LSP] : KI ratio was similar for both enzymes--6.6
for APase and 4.6 for PGase.
As was the case for enzyme activity, bacterial productivity
ranged widely. In the Ottawa River, values spanned three
orders of magnitude, peaking at 644 pmol C liter-’ d-l in
late spring with secondary peaks in late summer and early
autumn (Fig. 1). In the Maumee River, productivity oscillated with discharge and algal abundance. The highest value
measured was 17 1 vmol C liter-’ d-l. The pattern in the
Hudson River was more predictable-productivity
rose
steadily through spring, peaked in midsummer at 68 pmol
C liter-l d-l, and declined through autumn. The productivity
dynamics of these systems reflect their differences in hydrology and water chemistry. The Ottawa River drains a
small low-gradient watershed. The upper catchment is largely tile-drained agricultural land whereas the lower is predominantly suburban. Discharge and flow rates change
markedly with precipitation events, which also bring substantial quantities of nutrients, organic matter, and pollutants
Table 3. Correlation (Pearson r) bctwcen V,,, (the maximum
rate of enzymatic hydrolysis of artificial substrate) and apparent K,,,
(an indicator of the relative size of the labile substrate pool) for
Ottawa and Maumce River samples.
Enzyme
n
r
Fatty acid esterase
Leu-aminopcptidase
Alkaline phosphatase
p- 1,4-glucosidase
cy-1,4-glucosidase
Trypsin endopeptidase
54
56
55
54
51
46
0.76*
0.33*
-0.01
-0.16
0.01
0.39”
significant at (Y = 0.05.
0.3-2.5(1.1)
0.05-2.2(0.3)
0.02-2.4(0.5)
ND-O. 12(0.02)
ND-O. 17(0.03)
ND-O. 17(0.03)
ND-O. 16(0.03)
pXase
NAGase
* Statistically
Hudson
into the stream. Water quality varied widely, with high (40
mg liter I) inorganic nitrogen in the spring, significant riparian litter inputs in the autumn, and high salinity in the
late autumn (Table 1). The larger Maumee River is less influenced by riparian inputs and is consequently more “buffered” against changes in water chemistry and DOC quality.
The lower Hudson River is larger still and had the least
variation in productivity.
Not surprisingly, bacterial productivity was more closely
correlated with substrate hydrolysis rates (V,,,,,) than with
LSP size (K,,,). In the Ottawa and Maumee Rivers, only FAEase and TEPase K,,, values showed significant correlations
with bacterial productivity (r = 0.43 and 0.75, respectively).
In contrast, the V,,,,, values for all five commonly measured
enzymes were correlated with productivity: FAEase r =
0.44, LAPase r = 0.58, APase r = 0.39, cxGase r = 0.86,
PGase r = 0.86 (n = 66).
Another way to compare relationships between enzymatic
activity and production is to calculate, for each river, mean
productivity per unit enzyme activity. For FAEase, LAPase,
and TEPase, these dimensionless ratios were similar across
systems, -1, 4, and 1, respectively (Table 4). The ratios for
APase increased about 3-fold from the Hudson (1.8) to the
Ottawa (6.4) and from the Ottawa to the Maumee (16). For
aGase and PGase, the Maumee ratios were about 4-5 times
higher than those of the other two rivers. Collectively, productivity increments per unit enzyme activity increased from
the Ottawa to the Hudson to the Maumee River in a ratio of
1 : 1.7 : 7, presumably reflecting the trophic characteristics of
each system.
In terms of inorganic nutrient concentration, the Ottawa
River was the most eutrophic system (mean soluble reactive
P and inorganic N of -3.5 and 135 PM), the Hudson River
34
Sinsabaugh et al.
,700
r(
,
‘3-l
Table 4. Comparison of bacterial productivity (pm01 C liter-’
d ‘) per unit enzyme activity (pm01 substrate hydrolyzed liter-’ d- I)
in the Ottawa, Hudson, and Maumee Rivers.
Ottawa River
2 600
;rj
$500
“0
Productivity/activity
Enzyme
60
120
180
240
300
360
Fatty acid esterase
Leu-aminopeptidase
Alkaline phosphatase
,0-1,4-glucosidase
cy-1,4-glucosidase
Trypsin endopeptidase
P-N-acetylglucosaminidase
@xylosidase
Ottawa
Hudson
Maumee
1.o
3.5
6.4
23
17
1.1
*
*
1.1
3.5
1.8
30
49
*
34
35
1.1
4.4
16
141
202
0.6
*
*
* Activity not measured.
F 7o
8 60
.w
3
$
50
l$
20
Y
a
L2
k
0
10
60
120
180
240
300
- 200
“bl
@J
360
I
1 Maumee River
Z 160
$
60
120
180
240
300
360
Day of year
Fig. 1. Bacterioplankton productivity in the Ottawa and Hudson
Rivers from March through December 1994, and at two stations
(4, downstream; +, upstream) in the Maumee River from April to
December 1995.
the least (mean soluble reactive P and inorganic N of -1
and 70 PM), with the Maumee River intermediate (mean
soluble reactive P and inorganic N of -2.3 and 90 PM).
Bacterial productivity followed the same pattern (mean values 103, 78, 22 pmol C d-l liter-’ for the Ottawa, Maumee,
and Hudson, respectively) and so did esterase and peptidase
activity, as evidenced by the similarity of their productivity-
to-enzyme activity ratios. In terms of phytoplankton abundance, the Maumee River was the most autotrophic with a
mean chlclrophyll concentration of 1 mg liter-‘. Phosphatase
and carbohydrase activities per unit of production were low
compared to the Hudson and Ottawa Rivers, where phytoplankton abundances were two orders of magnitude lower
(mean chlorophyll 5 and IO ,ug liter-l, respectively). Thus,
it seems that in the Maumee River, bacterial substrate requirements for phosphorus and saccharides were subsidized
through algal metabolism. Taken collectively, the enzyme
and production trends suggest that bacterial productivity increases with inorganic nutrient concentration while the availability of algal exudates tends to reduce the level of enzyme
activity required to support production.
The enzyme and production data were used to evaluate
the MARCTE model for the functional organization of heterotrophic microbial communities. The C-acquiring enzyme
pool was represented by PGase and cxGase, the N-acquiring
enzyme pool by LAPase, and the P-acquiring enzyme pool
by APase. FAEase was not included in the EC pool because
the instability of MUF-acetate raised doubts about its reliability as a measure of lipase activity. Consistent with model
prediction., bacterial productivity was correlated with EC (r2
= 0.75) bl.tt not with ET (r2 = 0.02) (Fig. 2). (With FAEase
included in the EC pool, the r2 value reduced to 0.69.) Because most of the observations were clustered at low EC
values, the regression was strongly influenced by a small
subset of ,cases. Productivity was also positively correlated
with EC/E,, (r2 = 0.25) and EC/E, (9 = 0.36) (Fig. 3), suggesting that observed enzymatic patterns were due in part to
resource allocation among C-, N-, and P-acquiring enzymes.
The range of EC/E, values was about lo-fold greater than
the range of EC/EN values, but the slope of the EC/EN regression was lo-fold greater than that of the EC/E, regression.
Discussion
Like al I ecological processes, bacterial productivity is
subject to a complex hierarchy of regulatory constraints.
Large-scale relationships with reduced carbon abundance,
measured as bulk DOC or as algal productivity, have been
documented (Cole et al. 1988; Tranvik 1990). These rela-
Bacteria production
0.0
0.2
0.4
0.6
0.8
1.0
35
in rivers
0.0
0.2
E,/(l+ EN/E,+ E,/E,)
A 300
0.8
0.6
0.4
1.0
1.2
EC/EN
l
I
.%
b-
-:; 200 1
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0.0
2.0
4.0
6.0
8.0
10.0
ET
Fig. 2. Evaluation of the MARCIE model (upper) that describes
bacterial productivity as a function of the activity of C-acquiring
enzymes (y2 = 0.75, IZ = 66, F = 192). The lower plot is a null
model that assumes productivity is proportional to total extracellular
enzyme activity (r2 = 0.02, F = 1.5).
tionships are valuable for intersystem comparisons and for
making global predictions, but their resolution does not extend below the ecosystem level. At that level, productivity
reflects the local dynamics of macronutrient supply and community structure, neither of which can be easily characterized. However, if the problem is reduced further, to the cellular or population level, productivity can be described in
terms of the kinetics of substrate utilization (e.g. Characklis
and Marshall 1990).
For heterotrophic bacteria, substrate utilization is mediated by enzyme systems arrayed extracellularly. These systems include those involved in breaking covalent bonds (hydrolases and oxidases) and those involved in transporting
molecules across the cell membrane (permeases) (Price and
Morel 1990). The kinetics of both groups have been described with simple models, and comparative studies have
suggested that in natural systems it is most often the rate of
generation of utilizable molecules from polymeric or con-
Fig. 3. Evaluation of the resource allocation hypothesis of the
MARCIE model which predicts tradeoffs among the activities of
C-, N-, and P-acquiring enzymes (E,, EN,E,,, respectively). For the
upper panel, r 2 = 0.25, F = 21.6, n = 66, slope = 212, and intercept = 34. For the lower panel, r2 = 0.16, F = 12.1, slope = 21.5,
and intercept = 46 (if the outlying point in the upper left is excluded, r2 increases to 0.36).
densed substrates that limits bacterial metabolism (Chrost
1991).
The synthesis and secretion of ectoenzymes is presumably
expensive, especially in terms of nitrogen, and once deployed the cell has no control over their activity or turnover.
Drawing from many studies, there seems to be a common
hierarchic system of feedbacks that links ectoenzyme synthesis and activity with cellular metabolism to ensure the
efficient use of resources. When an adequate supply of directly utilizable substrates exists, the production of hydrolases is repressed at the level of transcription. However, even
under repression conditions, low-level, constitutive synthesis
is maintained. When the supply of utilizable carbon dwindles, the products of constitutive activity may, if sufficiently
abundant, induce the production of new enzymes by triggering a signal cascade that ultimately derepresses transcrip-
36
Sinsabaugh et al.
tion. Derepression results in the deployment of additional
enzyme until the point at which the assimilation of excess
reaction products triggers a repression feedback. This transcriptional control conserves resources but does not extend
beyond the cell.
Once deployed, the activity and persistence of ectoenzymes are subject to environmental control. One mode of
control is exerted by physicochemical variables such as temperature and pH. Another is effected by the relative abundance of substrates and inhibitors. Most enzymes are subject
to competitive inhibition by the endproducts of their activity,
which may accumulate if products are generated in excess
of cellular utilization. Depending on their deployment, ectoenzyme activity is also subject to noncompetitive inhibition. The most ubiquitous noncompetitive inhibitors are humic condensates, which typically affect enzymes by enhancing their stability (extending turnover time) and reducing
their activity (in Michaelis-Menton terms by increasing Km
and decreasing V,,,,,) (Burns 1983; Sinsabaugh and Linkins
1987; Lahdesmaki and Piispanen 1992).
Common regulatory mechanisms at the cellular and environmental levels suggest that predictable patterns of enzymatic activity may exist at the community level. The evidence consists of similarities in the abundances of various
enzymes in different systems, especially in relation to productivity or processes like decomposition, and correlations
between enzyme activities and nutrient supplies. As examples, Chr6st and Rai (1993) found that P-glucosidase activity
was correlated with bacterial production in nutrient-enriched
freshwater mesocosms, whereas alkaline phosphatase activity was linked to productivity in nutrient-impoverished ones.
Miller-Niklas
et al. (I 994) reported that cell-specific activities of cr-glucosidase, /3-glucosidase, Leu-aminopeptidase,
lipase, and chitinase were generally similar for bacteria associated with marine snow and for those suspended in the
surrounding seawater. They also found that these enzyme
activities were positively correlated with productivity, and
that rates of substrate hydrolysis and uptake were generally
similar. Correlations between substrate pool size, hydrolysis
rates, and product uptake were also made by Hoppe et al.
(1988) in a study of the utilization of dissolved combined
amino acids by nearshore bacterioplankton, and by Miinster
( 199 1) in a comparative study of glucose and amino acid
utilization by bacteria in eutrophic and polyhumic lakes. In
a broader survey, Kroer et al. (1994) found that specific Leuaminopeptidase activities were similar among marine, estuarine, and riverine bacterioplankton despite wide differences
in the concentration of inorganic N. Many studies, in both
aquatic and terrestrial systems, have documented inverse relationships between phosphorus availability and phosphatase
activity (e.g. Ammerman 1991; Cotner and Wetzel 1991;
Mulholland and Rosemond 1992); in some instances, this
activity has been used as an indicator of P availability (Wctzel 198 1; Gage and Gorham 1985). Decomposition rates for
particulate detritus have been correlated with the activity of
cellulose and lignin-degrading enzymes (Sinsabaugh et al.
1994a), and mass losses from marine snow have been linked
to high rates of hydrolytic activity (Smith et al. 1992). Collectively, these findings suggest a fundamental relationship
between emergent microbial processes and community enzyme aclivities.
The bIARCIE model is an attempt to synthesize these
patterns. The basic premise is that cellular ectoenzyme production and environmental nutrient limitation can be integrated at the community level through an optimal resource
allocation strategy that ties production, or decomposition,
rates directly to the distribution of enzyme activities. In the
model, enzyme activities serve as indicators of the relative
availability of carbon, nitrogen, and phosphorus. Although
originally developed to describe the relationship between
mass loss rates and microbial activity during plant litter decomposition (Sinsabaugh and Moorhead 1994), the MARCTE model may serve as a general model for heterotrophic
microbia‘l communities. In fact, it is similar to a conceptual
model for bacterioplankton presented by Chr6st and Rai
(1993). It also incorporates the suggestion of Sondergaard
and Middleboc (1995) that differences in ectoenzyme affinities (K,J among systems are responsible for the relationship
they describe between the labile fraction of DOC and total
DOC.
The principal advantage of the MARCIE model is that all
the indeplendent parameters are enzymatic activities, which
are easily measured and intimately linked to both substrate
pools ancl microbial metabolism. In some contexts, the model may offer an economical means of monitoring microbial
activity, as production or decomposition, and assessing relative nutrient limitations (Sinsabaugh et al. 1994b; Sinsabaugh and Findlay 1995; Jackson et al. 1995). However,
there are significant limitations and the model may prove to
be more heuristic than predictive.
In the Ottawa and Hudson Rivers, algal productivity was
low relative to that of bacteria and so were the ratios of
bacterial productivity to enzyme activity. In the more autotrophic Maumee River, the ratios were several times higher,
implying that bacterial production was supported by directly
utilizable algal exudates. Chr6st and Rai (1993) proposed
that the availability of monomeric substrates would likely
repress ectoenzyme synthesis, which would weaken correlations between hydrolase activity and production. This process may account in part for the clustering of most of our
data at low values of E,, even in cases where productivity
was moderately high.
We considered peptidases as N-acquiring enzymes, but
their relative importance as sources of carbon and nitrogen
may vary among systems (Kroer et al. 1994; Christian and
Karl 1995). Although we found positive relationships between productivity and E,IE, and EC/E,, which supported
the resour’ce allocation premise of the MARCIE model, there
is much about these relationships that remains obscure. First,
ratios of productivity to LAPase activity were similar in all
three systems, corroborating a similar observation by Kroer
et al. (19!)4). Thus, most of the increase in EJE, with productivity was due to increases in specific carbohydrase activities; E4Ease, like LAPase, showed little variation among
systems in relation to productivity. Ratios of productivity to
APase acl:ivity were more variable; thus, EC/E, spanned an
almost lCI-fold greater range than did EC/EN, but the slope
was only a 10th that of the EC/EN regression. Overall, our
results suggest that although bacterial productivity generally
Bacteria production
increases with inorganic nutrient availability, high levels of
productivity at any specific eutrophic state require a carbon
flow that includes both saccharides and amino acids, which
can be obtained either at enzymatic cost or through direct
algal subsidies. This interpretation is consistent with the
work of Kroer (1993) and Jorgensen et al. (1994) who have
shown that bacterial growth efficiency varies with substrate
C : N ratio.
The MARCIE model is an attempt to integrate heterotrophic nutrient acquisition activity over molecular, organismal,
and community scales. It provides an organizational paradigm for the large body of literature on enzymatic activity
in terrestrial and aquatic systems and a reference for further
studies. In the case of bacterioplankton, both the validations
and the deviations provide insight into the transformation of
detrital carbon.
References
1992. Standard methods
for the examination of water and wastewater, 19th ed. APHA.
AMMERMAN, J. W. 1991. Role of ecto-phosphohydrolases in phosphorus regeneration in estuarine and coastal ecosystems, p. 2559. Zn R. J. Chrost [ed.], Microbial enzymes in aquatic environments. Springer.
AZAM, E, AND OTHERS. 1983. The ecological role of water-column
microbes in the sea. Mar. Ecol. Prog. Ser. 10: 257-263.
BAKER, D. B. 1993. The Lake Erie Agroecosystem Program: Water
quality assessments. Agric. Ecosystems Environ, 46: 197-215.
BIDDANDA, B., S. OPSAHL, AND R. BENNER. 1994. Plankton respiration and carbon flux through bacterioplankton on the Louisiana shelf. Limnol. Oceanogr. 39: 1259-1275.
BoT’I’, T L., AND L. A. KAPLAN. 1990. Potential for protozoan
grazing of bacteria in streambed sediments. J. N. Am. Benthol.
AMERICAN PUBLIC HEALTM ASSOCIATION.
Sot. 9: 336-345:
-,
AND E T. KUSERK. 1984. Benthic bacterial biomass supported by streamwater dissolved organic matter. Micrab. Ecol. 10: 335-344.
BIJRNS, R. G. 1983. Extracellular enzyme-substrate interactions in
soil, p. 249-298. Zn R. H. Slater et al. [cds.], Microbes in their
natural environment. Cambridge.
CHARACKLIS, W. G., AND K. C. MARSIIALL [eds.]. 1990. Biofilms.
Wiley.
CHIIISTIAN, J. R., AND D. M. KARL. 1995. Bacterial ectocnzymes
in marine waters: Activity ratios and temperature responses in
three oceanographic provinces. Limnol. Oceanogr. 40: 10421049.
CHRIST, R. J. 1991. Environmental control of the synthesis and
activity of aquatic microbial ectoenzymes, p. 25-59. Zn R. J.
Chrcist [cd.], Microbial enzymes in aquatic cnvironmcnts.
Springer.
, AND H. RAI. 1993. Ectoenzyme activity and bacterial secondary production in nutrient-impoverished and nutrient-enriched freshwater mesocosms. Microb. Ecol. 25: 131-150.
COLE, J. J., N. E CARACO, AND B. L. Pcicrls. 1992. Can phytoplankton maintain a positive carbon balance in a turbid, freshwater, tidal estuary. Limnol. Oceanogr. 37: 16081617.
S. FINDI,AY, AND M. L. PACE. 1988. Bacterial production
in fresh and saltwater: A cross-system comparison. Mar. Ecol.
Prog. Ser. 43: l-10.
COTNER, J. B., AND R. G. WE’IZEI,. 1991. Bacterial phosphatascs
from different habitats in a small, hardwater lake, p. 187-205.
Zn R. J. Chrost red.], Microbial enzymes in aquatic environments. Springer.
in rivers
37
FINDI,AY, S., AND OTHERS. 1991. Weak coupling of bacterial and
algal production in a heterotrophic ecosystem, the Hudson estuary. Limnol. Oceanogr. 36: 268-278.
-,
M. L. PACE, D. LINTS, AND K. HOWE. 1992. Bacterial
metabolism of organic carbon in tidal freshwater Hudson estuary. Mar. Ecol. Prog. Ser. 89: 147-153.
GAGE M. A., AND E. GORHAM. 1985. Alkaline phosphatase activity
and cellular phosphorus as an index of the phosphorus status
of phytoplankton in Minnesota lakes. Freshwater Biol. 15:
227-233.
GONZALEZ, J. M., E. B. SHERR, AND B. E SIIERR.
1990. Sizeselective grazing on bacteria by natural assemblages of estuarine flagellates and ciliates. Appl. Environ. Microbial. 56: 583589.
HEDGES, J. I., AND OTHERS. 1994. Origins and processing of organic matter in the Amazon River as indicated by carbohydrates and amino acids. Limnol. Oceanogr. 39: 743-761.
HERNDL, G. J., G. MILLER-NIKI~AS,
AND J. FRICK. 1993. Major
role of ultraviolet-B in controlling bacterioplankton growth in
the surface layer of the ocean. Nature 361: 717-719.
HOLLIBAUGM, J. T., AND E AZAM. 1983. Microbial degradation of
dissolved proteins in seawater. Limnol. Oceanogr. 28: 11041116.
HOPPE, H. G. 1983. Significance of exoenzymatic activities in the
ecology of brackish water: Measurements by means of methylumbclliferyl substrates. Mar. Ecol. Prog. Ser. 11: 299-308.
S.-J. KIM, AND K. GOCKB. 1988. Microbial decomposition
in aquatic environments: Combined process of extracellular enzyme activity and substrate uptake. Appl. Environ. Microbial.
54: 784-790.
HUDSON, J. J., J. C. ROFF, AND B. K. BURNIS~N.
1992. Bacterial
productivity in forested and open streams in southern Ontario.
Can. J. Fish. Aquat. Sci. 49: 2412-2422.
JACKSON, C., C. FOREMAN, AND R. L. SINSABAUGM. 1995. Microbial enzyme activities as indicators of organic matter processing rates in a Lake Eric coastal wetland. Freshwater Biol. 34:
329-342.
JV)RGENSEN, N. 0. G., N. KROER, AND R. B. COFFIN. 1994. Utilization of dissolved nitrogen by heterotrophic bacterioplankton:
Effect of substrate C/N ratio. Appl. Environ, Microbial. 60:
4124-4133.
KAPLAN, L. A., AND T L. BOTT. 1983. Microbial heterotrophic
utilization of dissolved organic matter in a Piedmont stream.
Frcshwatcr Biol. 13: 363-377.
-,
AND -.
1985. Acclimation of stream-bed hcterotrophic microflora: Metabolic responses to dissolved organic matter. Freshwater Biol. 15: 479-492.
KROER, N. 1993. Bacterial growth efficiency on natural dissolved
organic matter. Limnol. Oceanogr. 38: 1282-1290.
-,
N. 0. G. JP)RGENSEN, AND R. B. COFFIN. 1994. Utilization
of dissolved nitrogen by heterotrophic bacterioplankton: A
comparison of three ecosystems. Appl. Environ. Microbial. 60:
4116-4123.
LAHDESMAKI, I?, AND R. PIISPANEN. 1992. Soil enzymology: Role
of protective colloid systems in the preservation of exoenzyme
activities in soil. Soil Boil. Biochem. 24: 1173-1177.
MEYER-REIL, L.-A. 1987. Seasonal and spatial distribution of extracellular enzymatic activities and microbial incorporation of
dissolved organic substrates in marine sediments. Appl. Environ. Microbial. 53: 1748-1755.
MORAN, M. A., AND R. E. HODSON. 1994. Dissolved humic substances of vascular plant origin in a coastal marinc environment. Limnol. Oceanogr. 39: 762-77 1.
MUIX)LLAND,
P J., AND A. D. ROSOMOND. 1992. Periphyton response to longitudinal nutrient depletion in a woodland stream.
J. N. Am. Benthol. Sot. 11: 405-419.
38
Sinsabaugh et al.
MILLER-NIKLAS,
G., S. SCHUSTER, E. KALTENB~CK, AND G. J.
HERNDL. 1994. Organic content and bacterial metabolism in
amorphous aggregations of the northern Adriatic Sea. Limnol.
Oceanogr. 39: 58-68.
MUNSTER, U. 1991. Extracellular enzyme activity in eutrophic and
polyhumic lakes, p, 25-59. Zn R. J. Chrost red.], Microbial
enzymes in aquatic environments. Springer.
POMEROY, L. R. 1974. The ocean’s food web, a changing paradigm.
Bioscience 24: 499-504.
PRICE, N. M., AND E M. M. MOREL. 1990. Role of extracellular
enzymatic reactions in natural waters, p. 235-258. Zn W.
Stumm red.], Aquatic chemical kinetics. Wiley.
RIHMANN, B., R. T. BELL, AND N. 0. G. JORGENSEN. 1990. Incorporation of thymidine, adenine and leucine into natural bacterial assemblages. Mar. Ecol. Prog. Ser. 65: 87-94.
SCAVIA, D., AND G. A. LAIKD. 1987. Bacterioplankton in Lake
Michigan: Dynamics, controls, and significance to carbon flux.
Limnol. Oceanogr. 32: 1017-1033.
SHIAH, E-K., AND H. W. DUCKLOW. 1994. Temperature regulation
of heterotrophic bacterioplankton abundance, production, and
specific growth rate in Chesapeake Bay. Limnol. Oceanogr. 39:
1243-l 258.
SINSABAUGH, R. L., AND S. FINIILAY. 1995. Microbial production,
enzyme activity and carbon turnover in surface sediments of
the Hudson River estuary. Microb. Ecol. 30: 127-141.
-,
AND A. E. LINKJNS. 1987. Inhibition of the Trichoderma
viride cellulase complex by leaf litter extracts. Soil Biol. Biothem. 19: 719-725.
-,
AND D. L. MOORHEAD. 1994. Resource allocation to extracellular enzyme production: A model for nitrogen and phosphorus control of litter decomposition. Soil Biol. Biochem. 26:
1305-133 1.
-
-AND A. E. LINKJNS. 1994a. The enzymic basis of
plant litter decomposition: emergence of an ecological process.
Appl. Soil Ecol. 1: 97-l 11.
M. OSGOOD, AND S. FINDLAY. 19946. Enzymatic models
fo; estimating decomposition rates of particulate detritus. J. N.
Am. Benthol. Sot. 13: 160-l 69.
SMJTH, D C., N. SIMON, A. L. ALI~IIREDGE, AND E AZAM. 1992.
Intensive hydrolytic enzyme activity on marine aggregates and
implications for rapid particle dissolution. Nature 359: 139142.
S~MEVJI,I.E, M., AND G. BILLEN. 1983. A method for determining
exoproteolytic activity in natural waters. Limnol. Oceanogr.
28: 190-l 9%
SONDERGL~ARD, M. AND M. MJDDLEBOE. 1995. A cross-system
analysis of labile dissolved organic carbon. Mar. Ecol. Prog.
Ser. 118: 283-294.
THURMAN, E. M. 1986. Organic geochemistry of natural waters.
Junk.
TRANVIK, L. 1990. Bacterioplankton growth on fractions of dissolved organic carbon of different molecular weights from humic and clear waters. Appl. Environ. Microbial. 56: 16721677.
WETZEL R.. G. 1981. Longterm dissolved and particulate alkaline
phosphatase activity in a hardwater lake in relation to lake
stability and phosphorus enrichments. Int. Ver. Theor. Angew.
Limnol. Verb, 21: 337-349.
WRIGHT, F!. T., AND J. E. H~~BIE. 1966. Use of glucose and acetate
by bacteria and algae in aquatic ecosystems. Ecology 47: 447453.
Submitted: 18 July 1995
Accepted: 21 April 1996
Amended: 28 August 1996