United States
Environmental Protection
Agency
Risk Reduction
Engineering Laboratory
Cincinnati OH 45268
Research and Development
EPA/600/S2-89/030 Feb. 1990
f/EPA Project Summary
Nutrients for Bacterial
Growth in Drinking Water:
Bioassay Evaluation
Louis A. Kaplan and Thomas L. Bott
The regrowth of bacteria in drinking
water distribution systems can lead
to the deterioration of water quality.
Pathogenic bacteria are heterotrophs
and heterotrophs are probably the
dominant bacteria associated with
the regrowth phenomenon. Only a
portion of the total organic carbon
(TOO) In drinking water is biologically
labile to heterotrophic bacteria, and a
bioassay developed to quantify this
assimilable organic carbon (AOC)
has been proposed as an index of the
regrowth potential of drinking water.
We have evaluated both biological
and chemical assays for determining
AOC as related to regrowth of bac-
teria in drinking waters from surface
water and groundwater sources.
Pseudomonas f/uorescens strain P-17
was used in bloassays for AOC.
Dissolved organic carbon (DOC), uv-
labile DOC, DOC < 10,000 daltons,
monosaccharides, and primary
amines were the chemical assays
used to predict concentrations of
AOC. Growth of P-17 was enum-
erated as viable and total cells with
spread plates and direct epifluores-
cence microscopy, respectively. AOC
concentrations in surface waters
ranged from 48 to 607 pg liter1 and in
a groundwater supply from 40 to 146
ng liter1. AOC remained relatively
constant or declined in distribution
systems with distance from the
source. Incubation vessel surface to
volume ratio influenced the AOC
value by enhancing wall growth of
reversibly attached cells. The bio-
assay assumes that (1) organic
carbon limits growth of the bioassay
organism, (2) the yield of the
bioassay organism on naturally
occurring AOC is constant and equal
to yield on model organic com-
pounds, and (3) the bioassay
organism is an appropriate surrogate
for the native microflora of distribu-
tion systems in utilizing AOC. We
have found that phosphorus addi-
tions to some test waters were
required to generate carbon limita-
tion and that yield on naturally
occurring AOC approximates the
yield on acetate. Correlations of the
bioassay AOC with chemical deter-
minations were poor, but with
improvements we have made in the
handling of the test water, glassware,
and P-17, we suggest that the
bioassay holds promise for a simple,
routine measure of drinking water
regrowth potential.
This Project Summary was devel-
oped by EPA's Risk Reduction Engi-
neering Laboratory, Cincinnati, OH, to
announce key findings of the research
project that Is fully documented In a
separate report of the same title (see
Project Report ordering information at
back).
Introduction
The regrowth of bacteria in drinking
water distribution systems can lead to the
deterioration of water quality and even
non-compliance of a supply. Bacterial
growth occurs on the walls of the
distribution system, and in the water
either as free living cells or cells attached
to suspended solids. Regrowth of
bacteria is a multi-faceted phenomenon
-------
influenced by temperature, residence
time in mains and storage units, the
efficacy of disinfection, and nutrients.
Secondary parameters probably include
redox potential, pH, inoculum size, shear
stress, main construction material, and
chlorine residual.
Regrowth within distribution systems
has largely been associated with
heterotrophic bacteria, organisms which
oxidize reduced carbon compounds for
energy and also require these organic
molecules as a source of carbon
"building blocks" for biosynthesis. Some
of those bacteria are opportunistic patho-
gens, and the growth of heterotrophs
may establish conditions conducive to
the growth of pathogenic bacteria.
Additionally, large numbers of hetero-
trophs have been shown to interfere with
the detection of coliforms, generate
anaerobic conditions conducive to cor-
rosion of pipe materials, and increase
chlorine demand.
Most of the energy and carbon for
heterotrophs in distribution systems pre-
sumably comes from dissolved organic
molecules in the source water, and
quantifying that nutrient supply was a
major focus of the research reported
here. The heterogeneous mixture of DOC
in groundwaters and surface waters used
for drinking water supplies ranges in
complexity from large molecules of
humic, fulvic, and hydrophilic acids, to
relatively simple compounds such as
carbohydrates, carboxylic acids, amino
acids, and hydrocarbons. It is the simple
compounds which are most susceptible
to microbial decomposition and they have
been collectively referred to as bio-
logically labile DOC in the ecological
literature.
The concept of labile DOC developed
in the ecological literature has been
applied to drinking water studies as the
AOC concept. The need to quantify AOC
has arisen, in part, because an easily
quantified chemical parameter used in
the drinking water industry, TOC, has not
been found to be a good predictor of
bacterial regrowth. The failure of TOC to
predict regrowth is understandable
because the ratio of AOC to TOC is not a
constant. However, AOC must be viewed
as one variable in a complex regrowth
equation, and the ability to quantify AOC
will be one step in the process of
understanding and eventually predicting
when and where regrowth will occur.
The research which is reported here
was undertaken as a cooperative agree-
ment with the U.S. EPA in an attempt to
develop a refined and field validated AOC
assay which could be used by water
utility technicians. The research included
an evaluation and modification of the
bioassay technique of D. van der Kooij
and associates of the Netherlands, and
an exploration of possible chemical
assays as an alternative to the bioassay.
We have also generated data which
describe changes in AOC between raw
and finished water, and AOC changes
during the transport of finished water
through two distribution systems.
Procedure
Preparation of Inoculum for AOC
Bioassay—Two different types of inocula
were used in these studies, log phase
cells and stationary phase cells of
Pseudomonas ffuorescens strain P-17.
An inoculum required to yield 1000 cells
mh1 was added with a sterile microliter
pipette to each bioassay vessel. Prior to
inoculation of a bioassay vessel with
P-17, a direct microscopic count was per-
formed to determine the appropriate
volume of inoculum.
Preparation of Incubation Water—
Finished drinking water used for bio-
assays was collected in organic carbon
free glassware containing 33.3 mg liter1
sodium thiosulfate and processed either
by pasteurization or filtration. Pasteur-
ization was performed by placing the
incubation vessels into a water bath
heated to 60 °C for 0.5 h.
Preparation of Bioassay Vessels and
Other Glassware—Glassware cleaning
involves a detergent wash, four rinses
with hot tap water, three rinses with 0.1N
HCI, four rinses with deionized water, and
heating to 550°C for 6 h. The pipettes
were treated in the same manner, except
that there was a 4 h cold tap water rinse
in a pipette washer between the deter-
gent wash and the hot water rinses.
Three different incubation vessels were
used, 1 liter Erlenmeyer flasks with
ground glass stoppers, BOD bottles, and
45 ml borosilicate vials with teflon lined
silicone septa. Commercially cleaned 45
ml borosilicate vials were used without
any prior treatment.
Enumeration of Bioassay Organism—
Incubation vessels were sampled using
asceptic technique and organic carbon
free sterile pipettes. Viable count
samples were placed into test tubes con-
taining sterile phosphate buffer and
serially diluted to 10'2, 10-3 and 10-". Re-
peated samplings over several days were
taken from the Erlenmeyer flasks and the
BOD bottles. The small size of the vials
allowed sufficient replication such that
each vial was sampled once, and
ferent vials were used to follow
change in population numbers over tir
Measurement of Cell Size and Ca
lation of Biovolume—Cell size was de
mined on formalin fixed samples stai
with acridine orange. P-17 is rod shaf
so the volume was calculated from
formula for a prolate spheroid, V = 4,
(L/2) (W/2)2. A minimum of twenty c
from a given bioassay vessel were rm
ured.
Determination of Cellular Carbon C
tent—Cells from incubation vessels w
filtered onto an organic carbon free gl
fiber filter (Gelman A/E). Filters w
oven dried at 60°C for 1 h £
combusted in a Carlo Erba Model 1
Elemental Analyzer* set up for carb
nitrogen, and hydrogen analysis. Divid
the amount of carbon on a filter by
number of cells per filter resulted in
estimate of carbon per cell.
Chemical Characterization of Asi
Water—All glassware used for orga
analyses was muffled at 550 °C to eli
inate organic carbon contamination. D<
was analyzed in either a Dohrmann
80 which used uv-promoted wet oxidat
or an Ol 700 which used persulfi
oxidation at elevated temperatur
(100°C). Low molecular weight DOC v»
determined by ultrafiltration through p
cleaned membranes in a 80 ml stirr
cell. Uv-labile DOC was determined b)
1 h uv irradiation of water samples held
quartz tubes without any additior
oxidant. DOC was measured before a
after the irradiation step, with t
difference being uv-labile DOC. Prime
amines were measured by a fluoromet
technique using fluorescamine, ai
monosaccharides were assayed using
spectrophotometric MBTH method.
Estimation of AOC Concentrations
Two different experimental approach
and designs were used to estimate AC
concentrations, (1) a bioassay bast
upon the work of van der Kooij, and (
the direct measurement of DOC uptal
by P-17. Both approaches require
growth of P-17 to stationary phase and i
estimation of cell numbers. The bioass<
approach further required an estimate
yield while the carbon uptake approai
required an estimate of carbon per cell.
The experimental design for the bi
assay approach initially involved sar
pling the incubation vessels on five da:
during stationary phase. Triplicate ve
'Mention of trade names or commercial product
does not constitute endorsement or recom
mendation for use
-------
sets were used for each of five acetate
ioncentrations. Each vessel was sub-
sampled in duplicate for both direct
microscopic counts and viable counts.
The subsamples for viable counts were
serially diluted, and a single spread plate
prepared for each dilution used. A single
filter was prepared from each subsample
used for direct microscopy. Beginning in
May of 1987, direct microscopic counts
were limited to a single sampling date
and the 0 and 250 pg C liter-1 acetate
concentrations, and after May of 1987,
viable counts were determined on 4
rather than the original 5 sampling dates.
In the bioassay approach, yield of P-17
on acetate was empirically derived for
viable and total cells through November
of 1987. Following that, only unamended
test water was used and the mean of all
previous yield values was applied to the
cell number data.
The direct measurement of DOC
changes in incubation vessels required
separating the organic carbon contained
in P-17 from solution without artifact.
Technical problems in accomplishing
sterile filtration without organic carbon
contamination necessitated the deter-
mination of TOG (including P-17 carbon)
and the calculation of DOC changes from
estimates of P-17 densities and carbon
per cell. The experimental design used
with this approach involved four replicate
vessels of unamended test water and
duplicate determinations of TOG concen-
trations in the vessels just prior to
inoculation and on the last sampling date
for stationary phase. Measurements of
organic carbon changes within the incu-
bation vessels and the densities of P-17
at stationary phase were used to
calculate the yield of P-17 on naturally
occurring AOC.
When BOD bottles were used as
incubation vessels, samples were poured
from the BOD bottles into vials for carbon
analysis. When vials were the incubation
vessels, replicate vials were sacrificed
after the heat fixation step for an initial
carbon analysis. An equivalent P-17
inoculum was added to carbon blanks to
verify that the inoculum was not a source
of measurable carbon. Controls for the
abiotic adsorption of TOC by the carbon
free glassware were carried through heat
treatment, but not inoculated.
Evaluation of P-17 as a Surrogate for
the Native Microflora of Water Dis-
tribution Systems—Sterile, carbon free
microscope slides were placed into a test
tube rack and suspended in a 100 liter
polyethylene tank continuously fed by
finished water from the distribution
system. After 4 months of incubation,
slides were removed and placed into a
rack submerged in site water. The slides
were transported back to our laboratory
and scraped with an organic carbon free
razor blade. The scrapings and rinses
were combined in a Corex tube, con-
centrated by centrifugation, sampled for
direct and viable cell enumeration, and
used as inocula in heat treated, de-
chlorinated test water.
Study Sites
Three different water supplies were
sampled during the course of this
investigation, the Chester Water Authority
reservoir on the Octoraro Creek in Lan-
caster County, PA, Well number 18 in
Chester County, PA from the Great
Valley Division of the Philadelphia Subur-
ban Water Company, and Pickering
Creek main pump station in Delaware
County, PA of the Philadelphia Suburban
Water Company (Figure 1).
Results and Discussion
Bacteriology of Water Used for AOC
Bioassays—None of the test water used
for bioassays had persistent coliform
problems. The Maximum Contaminant
Level established by the U.S. Environ-
mental Protection Agency is 1 coliform
(100 ml)-1 as a quality limit and 4 (100
ml)-1 as an action limit. The only detected
and verified coliforms in finished water
were associated with Pickering Creek
water in January when a taste and odor
problem occurred.
Chemistry of Drinking Waters Used for
AOC Bioassays—The correlation of or-
ganic constituents with AOC concentra-
tions was tested using data from 10
separate dates and 2 different water
sources. None of the correlations ex-
plained more than 16% of the variability
in the AOC data (Table 1).
Table 1. Correlation of Organic Constitu-
ents in Bioassay Water with AOC
Organic Constituent
Correlation
Coefficient
(r)
DOC
UV-Labile DOC
DOC < 10,000 Nominal
Molecular Weight
Primary Amines
Monosaccharides
0.397
0.268
0.032
0.162
0.387
Apparently the source of uv irradiation
used in this study was much more effec-
tive at oxidizing organic carbon than was
P-17, and the molecular weight cut-off of
10,000 daltons may have been too high
to have had much biological significance.
Primary amines and monosaccharides
can both be used by P-17, and the
carbon equivalents in these organic com-
pounds can be used as a check on AOC
determinations as minimum estimates,
but the plethora of additional carbon
molecules available in drinking water to
P-17 probably kept these classes of com-
pounds from being good predictors of
AOC concentrations.
Because simpler, faster chemical alter-
natives to the AOC bioassay were not
found, we proceeded to work on
validation of the bioassay. The bioassay
assumes that (1) organic carbon is
limiting to the growth of the bioassay
organism, (2) the yield of the bioassay
organism on naturally occurring AOC is
constant and equal to yield on model
organic compounds, and (3) the bioassay
organism is as capable as the native
microflora in a water distribution system
in utilizing AOC.
Influence of Incubation Vessel Size—
The 1 liter Erlenmeyer flasks with ground
glass stoppers are relatively expensive
and awkward to handle. BOD bottles and
eventually 45 ml vials were substituted
for the flasks. Changing the size of the
incubation vessel did not diminish the
precision of the assay, as shown in Table
2. In Table 2, the replicate spread plate
variation represents duplicate spread
plates prepared from a single dilution
tube. This was performed as a quality
control measure on approximately 10%
of the spread plates. The estimate of
within vessel variation was derived from
all duplicate samples from vessels, iden-
tified by level of acetate addition, vessel
replicate, and sample date. The between
vessel variation was derived from the
triplicate vessels, separated by level of
acetate addition and sample date, and
the between incubation day variation was
derived from the estimates of P-17
densities at stationary phase over several
days, separated by site, initial sample
date, vessel type, and level of acetate
addition.
Although the number of data for the
vials are limited, the current data indicate
that within vessel, between vessel, and
between sampling day variation were
unaffected by reducing the size of the
incubation vessel. Several benefits were
gained by using the vials, especially the
simplification of glassware preparation.
The vials are inexpensive, so different
vials can be used for each sampling
-------
A
N
Pennsylvania
T*
Well 18
North Wayne
Philadelphia Suburban Water Co.
Pickering Creek Treatment
Plant/Pump Station
Bryn Mawr
Chester Water Authority
Rt. 52
Octoraro Raw Water I Village Green
Reservoir \ Octoraro Treatment Plant!
Pump Station
48" Transmission
Main
Ithan
Oakmont
Scale 1" = 9 Miles
Figure 1. Map of study sites.
Table 2. Estimates of Precision for Different Vessels Used in the AOC Bioassay
Coefficient of Variation"
Vessel Type
Flasks
BOD Bottles
Vials
Replicate
Spread Plates
9.9 ± 6.7 (150)
12.1 ± 10.5(102)
Within Vessel
Variation
11.2 ± 9.0(280)
10.8 ± 9.9 (821)
25.0 i 19.1 (24)
Between Vessel
Variation
13.7 ± 9.5 (95)
15.6 ± 11.7 (249)
13.4 (1)
Between Incubation
Day Variation
11.4 + 7.7(20)
15.7 ± 7.5 (62)
"Coefficient of variation is calculated from (x ± SO) 100, and the data are expressed as x ± SO (n).
during stationary phase. This avoids
repeated sampling of the same bioassay
vessel over a period of days, eliminating
the need for organic carbon free pipettes
and reducing the risk of carbon or
bacterial contamination. Preliminary tests
with commercially available precleaned
vials and teflon faced silicone septa show
less than 20 u9 C liter1 of contamination.
The use of the smaller vessels with
higher surface to volume ratios yielded
higher densities of P-17 in stationary
phase (Figure 2). The differences be-
tween cell densities in the vials, BOD
bottles, and the flasks are believed to be
related to wall effects. The dimensions of
the vials, BOD bottles, and flasks are
such that the surface to volume ratio of
the vial is twice that of the BOO bottle
and three times that of the flask. The
benefits to bacteria of association with
surfaces in nutrient poor environments,
i.e. higher concentration of nutrients, was
demonstrated for E. coli using glass
beads 40 years ago and it appears that
increased surface area is advantageous
for P-17 under our culture conditions.
Comparison of Viable and Total Cell
Enumerations—In general the density of
total cells, determined by direct micro-
scopy, exceeded the density of viable
cells determined as colony forming units,
by a factor of 1.6 (1.6 ± 0.6 (39) x ± S
(n) (Figure 3)). This relationship seem
reasonable in that P-17 is easily culture
on nutrient agar and the population i
stationary phase would be expected t
contain a significant proportion of dea
cells.
Cell Yield—The influence of all enurr
erating techniques on yield is shown i
Figure 4. Assuming carbon limitation, th
yield of P-17 on acetate should not var
between test waters, and there are n
reasons to expect variations in yieh
based upon viable or direct enumeratio
of P-17 or vessel size. Phosphorus limita
tion was observed only at elevated level
-------
20—1
15 —
•S
6 «M
1 ,
0 —
I
10
I
t5
\
20
Figure 2.
Maximum Density in Larger Vessels
(CFUmr1 * 10s)
Influence of vessel size on densities of P-17: (m) Flask versus BOD bottle
(*) BOD bottle versus vial comparisons.
of organic carbon when acetate was
added to incubation vessels to measure
yield. The assumption of carbon limitation
was probably valid in all unamended test
waters. Our empirically derived estimates
of yield for the entire data set were
similar for viable and total cells averaging
3.24 ± 0.41 x 106 cfu dig acetate-C)-1
and 3.68 ± 0.85 x 106 cell dig acetate-
C)-1 (x ± SD (n = 20)), respectively.
Where yield was determined for more
than one vessel type, cellular yield was
found to be independent of vessel size. In
those instances, the average yields
based upon viable counts were 3.15 ±
0.30 x 106 and 3.15 ± 0.42 x 1Q6 (x ±
SD (n = 7)), and based upon total cells
were 3.67 ± 0.86 x 106 and 3.53 ± 1.46
x 106 (x ± SD (n = 6)) for flasks and
BOD bottles, respectively. We also used
the estimates of cellular carbon in P-17
determined with the elemental analyzer
(1.73 ± 0.03 x 10-7 ug C cell-1 (n = 4))
to convert the cellular yield on acetate to
cellular-C yield:
(1.73x10-7^0 cell-1)
(3.24 x 106 cfu (pg acetate-C)-1)
= 0.56 ng cell-C fog acetate-C)-1
Theoretical yields of heterotrophic bac-
teria on acetate have been calculated
from stoichiometry. With ammonia as the
nitrogen source, the theoretical yield on
acetate was 0.36, and when nitrate was
the nitrogen source, 0.28. These values,
however, are expressed as g cells (g
substrate)-1 and can be converted to
units of C assuming that 50% of cell dry
weight is C, and the knowledge that 40%
of acetate is C. The adjusted values then
become 0.45 and 0.38, respectively.
AOC Concentrations in Field Sam-
p/es—Unlike the estimates of P-17 yield
on acetate-C, the estimates of AOC
concentration are very dependent upon
both P-17 enumeration technique and
vessel size. The influence of enumeration
technique is fairly obvious. Direct enum-
eration of total cells, as discussed above,
gives higher values for maximum cell
densities which will translate into higher
AOC concentrations when divided by a
constant yield factor. The influence of
vessel type is harder to explain, but
probably involves wall effects as dis-
cussed above. The impact of these find-
ings on the AOC bioassay is that they
clearly require the assay to be operation-
ally defined as to vessel size and enum-
eration technique if comparable data are
to be generated.
Keeping those caveats in mind, the
AOC values measured in this study
ranged from 48 to 607 jig C liter1 and
were generally lowest for the Well 18
water, intermediate for Chester Water
Authority, and highest for Philadelphia
Suburban's Pickering Creek distribution
system. No seasonal patterns were
apparent in the surface water supplies,
but there is some evidence of biological
stabilization of the water within the
distribution systems (Figure 5). This was
most noticeable in the Pickering Creek
system, especially in the January and
February samples. In the Chester Water
Authority system and the Pickering
Creek system during the April sample,
little change was observed in AOC
beyond the raw water.
Yield on Naturally Occurring AOC—An
important assumption in the AOC bio-
assay is that the yield of P-17 on
acetate-C is equivalent to the yield of
P-17 on naturally occurring AOC. Pre-
vious studies with P-17 have shown that
the yield of P-17 on amino acids,
carboxylic acids, carbohydrates, and
aromatic acids ranged from 3.2 to 7.8 x
106 cells (ng carbon)-1. Our approach to
determining the yield on naturally
occurring AOC was to perform carbon
mass balance measurements in the
incubation vessels. Rather than separate
the P-17 cells from stationary phase
cultures for the measurement of organic
carbon uptake, we measured the TOC in
the incubation water, thus including
cellular carbon from the bioassay
organisms present in the water. We also
measured the carbon content of P-17
cells and applied that conversion factor to
the calculation of carbon changes during
incubation. The data presented in Figure
6 represent values which assume
complete oxidation of the cellular carbon
present during the TOC analyses. AOC
concentrations based upon these
separate estimation techniques were very
similar, implying that the yield of P-17 on
acetate-C is a reasonable approximation
of the yield on naturally occurring AOC.
Distribution System Microflora—A final
question addressed in this study was the
suitability of P-17 as a surrogate for
heterotrophic bacteria within distribution
systems. The initial enumeration of inoc-
ula used in this comparison indicated that
the density of bacteria on the slides
incubated for 4 months at the Bryn Mawr
site was 6x10" cell cm-2, and that only
-------
25—1
20 —
75-1
-
f 8
s
Q
5 —
0 —
r
0
I I \ \
5 10 15 20
Density from Viable Counts
(CPU ml'1 x 10s)
I
25
Figure 3. Influence of cell enumeration technique on densities of P-17.
7.5—1
5.0 —
Q 7
M =
-------
600-i
s
O J 300H
«? O
o-
U-09-88
Ef-07-88
Pickering
N. Wayne
Ithan
1 I I
0/0 13.4/3.4 19.3/5 4
Distance/Time of Travel
(km/h)
Figure 5. AOC concentrations in a water distribution system.
9. No seasonal patterns in surface
water AOC concentrations were ap-
parent, but there was evidence of
biological stabilization of water within
a distribution system.
Recommendations
*\. The simplified AOC bioassay devel-
oped as part of this research project
needs to be tested with a larger
number of water types and users.
2. The yield of P-17 on naturally occur-
ring AOC needs to be measured in a
greater number of water types.
3. Any AOC bioassay needs to be
standardized to vessel type and
method of organism enumeration.
4. The AOC bioassay should be per-
Oakmont formed over at least three separate
days and criteria developed for ac-
ceptance or rejection of the bioassay
results.
5. More research is needed as to the
efficacy of P-17 versus the native
microflora of water distribution sys-
tems in the utilization of and subse-
quent growth on AOC.
I
27.5/7.7
600—i
400 —
05
s:
o
E a
o
2 200 —
0 —
I
200
I
400
600
AOC from C Mass Balance
(pg-C Liter'')
Figure 6. AOC concentrations estimated from P-17 density and yield compared to
estimated from carbon mass balance: (m) BOD bottles; (•) vials.
-------
Louis A. Kaplan and Thomas L Bott are with the Academy of Natural Sciences of
Philadelphia, Avondale, PA 19311.
Donald J. Reasoner is the EPA Project Officer (see below).
The complete report, entitled "Nutrients for Bacterial Growth in Drinking Water:
Bioassay Evaluation," (Order No. PB 89-213 995/AS; Cost: $15.95, subject to
change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Risk Reduction Engineering Laboratory
U.S. Environmental Protection Agency
Cincinnati, OH 45268
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Official Business
Penalty for Private Use $300
EPA/600/S2-89/030
------- |