EVALUATION OF TOTAL COLIFORM, FECAL COLIFORM, AND
   FECAL STREPTOCOCCI AS ADEQUATE INDICATORS IN
  MONITORING PUBLIC WATER SUPPLY QUALITY IN THE
                         TROPICS
                            by
                   Jose Marcos Soto  Munoz
        A Thesis  submitted to the  Department  of BiologN

               FACULTY OF NATURAL SCIENCES
                 UNIVERSITY OF PUERTO RJCO
                   RIO PDEDRAS CAMPUS
           in  partial fulfillment of the  requirements
                     for  the degree of
               MASTER OF SCIENCE IN BIOLOGY
                      February,  1989
                  Ri'o Piedras, Puerto Rico

-------
    This thesis has been  accepted by the  faculty of the;
               DEPARTMENT OF BIOLOGY
            FACULTY OF NATURAL SCIENCES
              UNIVERSITY OF PUERTO RICO
                 RIO PIEDRAS CAMPUS
in  partial  fulfillment of the requirements  for the degree  of
            MASTER OF SCIENCE IN BIOLOGY
                   Thesis Committee:
                                                     Advisor
                               Terry C. Hazen, Ph. D.
                           \    Gary A. Toranzos^
                              Ivette Garcfa Castro, Ph. D.
                                            __
                            __,
                                Sharon K.  File, Ph.  D.

-------
                              DEDICATION

When  I  want to discover something, I  begin by  reading  up everything
that  has  been done  along  that line in the past  -  that's  what all the books
in the library arc for.   I  see what has  been accomplished  at great  labor
and  expense  in  the  past.  I gather the  data of many experiments as a
starting point and then  I  make many more.  The three essentials to
achieve  anything  worthwhile are,  first,  hard  work;  second,  stick-to-it-
iveness;  third, common sense.
                                                    -Thomas A.   Edison
                 To those  who  have thought  they can't.

-------
                         ACKNOWDLEGMENTS
      It is  impossible to properly  thank all the people  that helped  and
 supported me during  the course of this project in so little  space.  To all
 of you, thanks.   You  know  who you  are.
      My parents,  Jose  Luis and Carmen Ana, were a constant  source  of
 support.  So  were the rest  of  my  family: my grandmother, Marcolina;
 my  brothers,  Jor;i Luis and Juan  Sandalio; my sister-in-law, Doris; and
 my  niece,  Carla.
      I thank  Dr.   Ivette  Garcia-Castro, Dr.  Sharon K.  File, and Dr.  Gary
 Toranzos  for  their contributions  as  part  of  my  thesis committee.
      My good friends  from the  graduate  program, especially  Maria
 Socorro Flores,  Patricia Marcos (and  John Calderon),  and Maggie Rivera
 provided  much  needed  encouragement  and  good laughs.  Although they
 were not  physically close, Ivette Lopez and Tom  Abbruscato gave  me
 not  only  their  friendship, but  also  their support through  letters, phone
 calls, and occasional  visits.  The  staff at  the Microbial Ecology
 Laboratory,  especially Luz E.  Rodriguez, Madeline Bermudez,  Ismael
 Perez, and  Yazmfn Rojas, always provided  useful suggestions,  technical
 advice,  cheerfulness,  and friendly  help.
      I  thank  all at EPA's Caribbean  Field  Office,  especially Racqueline
 Shelton, Jorge Martinez and  the  rest of the Water  Management  Staff,  for
 their unconditional support  and  understanding.
      My  deepest gratitude goes to  my  advisor, Dr.  Terry C.  Hazen. for
giving me  this  opportunity  and then guiding me across great  distance.

-------
      though we  were far  apart, he was  always available for  questioning
and made many useful  comments  and suggestions.   Thanks  a  lot!
      This  research was  supported in part  by  the U.  S.   Environmental
Protection Agency,  Water Management Division;  Region II, Caribbean
Fkid Office.

-------
                           TABLE OF CONTENTS

                                                            Page No.

Dedication		„..,...	„	..			ii
Acknowdlegments	.iii
Table  of Contents	_	„	~-	„	v
List of Tables...,	vi
List of Figures	viii
Abstract	ix
Introduction	1
Materials and Methods	1 7
Results	27
Discussion	_	„	3 6
Conclusions	„	5 7
Literature  Cited	„	5 9
Tables	72
Figures	9 2
Appendix  I	! 0 2
Appendix  II	1 0 3
Appendix   III	1 0 4
Appendix  IV .„	-	-	,	1 0 5
Appendix  V		_			_	~	1 0 6

-------
                             LIST OF TABLES
                                                                 Page No.

f$ble  1.   Water Quality Parameter  Averages 	7 2

Table  2.   Bacterial  Density  Averages 	.-*	7 3

Table  3.  Direct Count and Percent Activity Averages	74

Table  4.   Correlation Matrix: Combined  Influent  Bacteriological
          Methods and Water Quality	7 5

Table  5.   Correlation Matrix: Bayamon Influent   Bacteriological
          Methods and Water Quality	7 6

Table  6.   Correlation Matrix: Bayamon Effluent  Bacteriological
          Methods and Water Quality	.........7 7

Table  7.   Correlation Matrix: Guaynabo  Influent  Bacteriological
          Methods and Water Quality	7 8

Table  8.   Correlation Matrix: Guaynabo Effluent Bacteriological
          Methods and Water Quality	7 9

Table  9.   Correlation Matrix: Villalba  Influent Bacteriological
          Methods and Water Quality	8 0

Table  10.   Correlation  Matrix:  Villalba Effluent Bacteriological
          Methods and Water Quality	8 1

Table  11.   Correlation  Matrix:  Yauco  Influent  Bacteriological
          Methods and Water Quality	8 2

Table  12.  Correlation Matrix: Yauco Effluent  Bacteriological
          Methods and Water Quality	8 3

-------
Table  13.   Media Performance in the Assessment of Target
          Bacteria.	
Table  14.   Media Performance in the Assessment of Nomarget
          Bacteria.								....	85

Table  15.   Specificity  and Selectivity of Viable Count Methods In the
          Assessment  of Influents	8 6

Table  16.   Specificity  and Selectivity of Viable Count Methods in the
          Assessment of  Effluents	8 7

Table  17.   Average  indicator Organism / AODC Ratios per Viable
          Count Method Assessed	8 8

Table  18.   Bacterial Species  Isolated  from Influent  Waters	89

Table  19.   Bacterial  Species Isolated  from Effluent Waters	90

Table  20.   Bacterial  Species Isolated  for  Chlorinated  Effluents	9 1

-------
                             LIST OF FIGURES
                                                                      Page  No

Figure  1.   Study Sites	;.._	9 2

Figure  2.   Total Coliform Densities  found in Bayamon  Influent
           Waters	93

Figure  3.   Total Coliform Densities  found in Guaynabo Influent
           Waters	94

Figure  4.   Total  Coliform  Densities found in Yauco Influent Waters  ....95

Figure  5.   Total  Coliform Densities  found in  Villalba Influent
           Waters	96

Figure  6.   Fecal Coliform Densities  found in Bayamon  Influent
           Waters	_	.97

Figure  7.   Fecal  Coliform Densities  found in  Guaynabo Influent
           Waters	98

Figure  8.   Fecal  Coliform  Densities  found in  Villalba Influent
           Waters	99

Figure  9.   Fecal  Coliform  Densities found in Yauco Influent
           Waters	,	1  0 0

Figure  10.  Fecal Streptococci  Densid.es found in  Villalba
             Effluent Waters	1  0 1

Figure 11.   Influent AODC at all  Sites 	102

-------
    EVALUATION OF TOTAL COLIFORM, FECAL COLIFORM, AND
       FECAL STREPTOCOCCI AS ADEQUATE  INDICATORS IN
      MONITORING PUBLIC  WATER SUPPLY QUALITY IN THE
                              TROPICS,

 JOSE MARCOS SOTO MUNOZ

 ADVISOR: DR. TERRY C HAZEN

                              ABSTRACT

      Previous studies  in  tropical  areas have  demonstrated that  high
 false positive  and negative  errors were  commonplace  when  determining
 whether  fecaJ contamination  of source  water has  occurred.   Other
 studies  had  shown  that microbial  pathogens can  have  higher  survival
 rates relative to  standard  indicator bacteria.   These violations  to  the
 criteria  used  to select  an  ideaJ fecal contamination indicator  organism
 proposed  doubts  about the use of standard  techniques in determining
 the  quality  of tropical  public  water supply systems.
      To determine if Escherichia coij  was  an  adequate indicator of
 tropical drinking  water quality, influents and  effluents from  four water
 treatment  plants in Puerto  Rico were  tested  weekly for i three  month
 period.   Membrane  filtration  and  multiple rube  fermentation  techniques
 were  used  to  detect  total conforms, fecal coliforms, and fecal
streptococci on appropiate  selective media.   All  presumptive positive
results were  confirmed  according  to standard  methodology.   From  each
sample,  an  appreciate  percentage  of  organisms giving  confirmed fecal

-------
coliform  results were  isolated and  identified.   Total  bacterial densities
were determined by  acridine  orange direct  counts.
      Results  show  that while  plant effluents  generally complied  with
current federal regulations, high  densities of bacteria  were present in
drinking  water.   Among these  bacteria,  injured  (viable but
unculmrabl*)  conforms were detected by  m-T7  agar.   High  levels of
false positive  results were  found for both the raw and drinking  water
samples.   Turbidity levels  apparently  affected  the disinfection process,
the  normal expression of indicator  bacteria  in  standard media, and  the
methodologies  used  to assess  water quality.   Neither  coliforms nor fecal
streptococci  were  adequate  indicators  of  tropical  water quality.

-------
                            INTRODUCTION

      In  recent years,  the  number  of cases  of waterborne diseases in
Puerto  Rico has increased  every y«^r (Puerto Rico Department  of Health,
1986).   The most  probable  cause for this, according to the  United States
Environmental  Protection  Agency (EPA),  the Environmental Quality
Board (EQB) and  the Puerto Rico  Acqueduct and  Sewer Authority
(PRASA), is  the over capacity  and antiquity  of the island's   water
treatment  and purification  plants, as well as  the  increased  quantities  of
raw sewage and  other  pollutants  being discharged into  the  island's
water  supplies  (Ruiz de la Mata, 1985).   The mixture  of pathogenic
bacteria and  industrial  pollutants,  along  with the possible
ineffectiveness  of the  water  treatment  process  represent an increased
risk of  waterborne disease epidemics.
      The standard techniques  being used to assess the microbiological
quality of  treated  water (coliform  bacteria  assays) were developed  for
use  in  temperate climates,  where  the  target indicator  organism,
Escherichia. coii.  is thermotolerant,  whereas the  environmental  bacterial
flora is  not.  In the tropics, most  assumptions  about  the indicator
organism  become  invalid  since some environmental flora can  withstand
the  higher  temperatures used to  recover  E. coli.  thus  resulting  in  high
numbers of  false positives (Santiago- Mercado and Hazen,  1987).
Studies made in the  Microbial  Ecology Laboratory of The University  of
Puerto Rico have  shown  an increased  pathogen survival rate relative to
regrowth  of the indicator  organism  in  the  environment (L6pez-Torres,
1982; Hazen et  al.,  1982;  Biam6n and Hazen, 1983).

-------
      Accurate  determination  of the  level  of  biological contamination  is
extremely important in  tropical  areas,  since  these  areas  possess  larger
numbers  of  waterborne  diseases.   This determination is more difficult in
tropical  source water  than  it is in temperate  waters due-  to several
factors.   Among the  physico-chemical  variables that affect  accurate
determinations of biological pollution in the tropics, we find
temperature,  high  organic content,  high light  intensity, and  heavy
rainfall  (Hazen,  1988).
      In  temperate  climates,  water  temperature  ranges from freezing  to
30°C, while  tropical water  temperature  may be as high as  45°  and never
goes  below  15°C (Hill and  Rai, 1982).  In  the  tropical  forests of Puerto
Rico,  water  temperature  ranged between  18°  and  24CC  throughout the
year  (Carrillo et ah, 1985;  Ldpez et  ah, 1987).  This thermal environment
results  in permanently stratified  (oligomictic)  water reservoirs  as well
as  lower amounts of  dissolved oxygen relative  to  temperate waters
(Hutchinson,  1977).  In  Puerto  Rico,  large  diurnal  variations (1  to  8
mg/L) in the dissolved  oxygen content of  water have  been  measured
(Ldpez-Torres et al., 1987;  Carrillo et al., 1985;  Perez-Rosas and Hazen,
1988).   This indicates  that tropical  source  waters  become anoxic  faster
than  those  of temperate  climates,  and  thus, few  organisms  apart  from
mesophilic,  facultatively  anaerobic bacteria can  thrive  in  the
environment (Hazen, 1988;  Carrillo et al.,  1985).
      Light  intensity in  the tropics  can adversely  affect  our perception
of  the levels of  indicator bacteria  present in tropical source water.   The
constantly high  light  intensity, along  with  high  temperatures, cause
constant   hypereutrophy of  tropical  surface  waters.   The  higher densities
of  naturally  occurring  bacteria  in  tropical  waters can  limit  the reliability

-------
 of   assays  that employ viable  counts  to  determine the  level  of  biological
 water contamination.   Tropical source waters have  a tendency  to
 overgrow on standard media  (Lo'pez-Torres et al.,  1987;  Samiago-
 Mercado and Hazen,  1987).   Also, non-indicator (background)  bacteria
 can  produce  bacteriocins  that  inhibit  the  growth  or  the  typical
 appearance  of indicator  bacteria on standard  media  (Means  and Olson,
 1981; Burlingame  et  al.,  1986).
      Tropical rainfall also affects dramatically  the  microbial content of
 water.   Some watersheds  in Puerto  Rico may receive more  than  10 cm
 of rain in 24 h (Carrillo et al., 1985).  Heavy rainfall changes the
 nutrient and  microbial  content of  water reservoirs  by  flushing  the
 surrounding  land  and vegetation into  the  reservoir (Carrillo  et al.,  1985;
 Hill and Rai, 1982; Oluwande et  al., 1983).  The washing  of soil into
 streams  and  lakes  also  increases the  turbidity of  reservoirs.   Studies in
 Africa, Puerto Rico,  and  Hawaii have clearly demonstrated that
 increases in  total  bacterial densities  coincide with  increasing rainfall
 (Barrel and Roland, 1979;  Oluwande et al., 1983;  Carrillo et al., 1985;
 Fujioka  and  Shizumura, 1985).
      When  discussing  the variables that  make  tropical  water quality  so
 different  from that of temperate water, we must  also  take  into
 consideration the  fact that  the quantity and quality  of bacterial
 resources are different  between  these  two  environments.   The
 hypereutrophic state of  tropicaJ  waters,  combined with  the higher
 temperatures, create a  high organic  content  environment  dominated by
 mesophilic and  thermotolerant  microbes (Hill  and Rai,  1982; Santiago-
Mercado and  Hazen, 1987).  This microflora  is similar  to  that of the gut
of animals (Hazen,  1988).   Tropical waters also  have higher  numbers of

-------
both  autochtonous and allochtonous microbes  (Hill and  Rai,  1982).   The
combination  of the particular -microbial  population and  environmental
conditions  of tropical  source waters  result in microbial survival  and
activity  rates that  are  totally different from those of  temperate waters
(Hazen,  1988;  McFeters and Stewart, 1972).
       In  the  tropics there are a  lot  more pathogenic microorganisms
than  there  are  in  temperate  climate  zones.   In fact, many tropical
diseases  are  totally unknown  in  temperate climates.   Therefore,
microorganisms  to  be  used  as  indicators of  tropical  water contamination
must cover a wider range and  diversity of pathogens.  However,  tropical
nations  have  adopted  regulations developed for  industrialized
temperate nations without testing their validity,  or the  validity of
techniques  used  to  assess water  quality  under these  regulations, in  their
particular  circumstances.   This  often  results in these  nations not being
able  to  meet their  own  water quality standards (Hazen,  1988).   This
problem arises  in  part from  a number of social  and economic factors.
Although 65%  of the world's population lives  in  the  tropics,  these  people
actually  have less than  10% of the the world's wealth (Odum,  1971).
Since  resources  are  limited,  funding  for the costly research  needed  to
find  better indicators of water pollution are  virtually  non-existent.
      False estimates of contamination  can be avoided  by studying  the
methods currently  being  used for water  analysis  in Puerto Rico as well
as  the environmental flora present  in source water and in finished
drinking water.   Studies of  this  nature would reduce the possibilities  of
waterborne  disease  epidemics  by  allowing us to understand better what
bacterial contamination indicators  are telling us.    Results of  this study

-------
could  be used  to  modify existing  regulations regarding  the  bacterial
quality of finished drinking  water  in  tropical climates.
STATEMENT OF PROBLEM:
      Tropical waters have  been identified as carriers of a  large  number
of  pathogenic  microorganisms.  In Puerto Rico,  water  reservoirs  receive
large quantities  of  industrial,  agricultural,  and domestic  sewage
effluents which,  when  combined  with  the potentially pathogenic
microbes in the  environment, represent  an  increased risk  of waterborne
disease  outbreaks.  The techniques commonly  used to  examine  the
microbiological quality  of drinking water rely on  assumptions  made
about  the  target indicator organism (Escherichia coin,  especially
thermotolerance  (Bonde, 1977).   These  assumptions may be  invalid  in
tropical  climates.   Regrowth  of  the indicator  organism, as well as  lower
pathogen survival  rates have  been  observed  in tropical  areas (Carrillo et
al.,  1985; Lopez-Torres  et al.,  1987; Hagler et  al., 1986; Fujioka  et  al.,
1981;  Lavoie,  1983; Hazen et  al.,  1982;  Hazen et  al., 1987;  Santiago-
Mercado  and Hazen,  1987).
      Water  quality in Puerto  Rico  is a  major problem.   Various types of
effluent,  such  as  domestic  and  municipal sewage,  industrial  effluents
and agricultural  effluents and run-offs,  are  being  discharged into most
of  the island's rivers  and coastal areas.   In  1978, the United States
Water  Resources Council reported that %%  of ail  samples  collected  from
24  sampling stations  throughout the  island  violated  the total coliform
standards.   The  United  Slates Geological  Survey  has since reported  that
80.6%  of all water sampling stations  in  Puerto Rico violated  the
recommended  maximum contaminant  levels   for recreational  waters
(<1000  fecal coliforms / 100 ml) (Curtis  et al., 1984).  As  contamination

-------
of  public water supplies  increases,  so  has  the number of cases of
gastroenteritis  (possibly  waterborne)  (Puerto  Rico Department of Health,
1985).   In  1976, improper chlorination of  the drinking water supply  in
the town of Comerfo  resulted  in 7,800 cases of gastroenteritis  (U.  S.
Water Resources Council, 1978).   During  the  summer of 1987, an
outbreak of gastroenteritis and  dysentery  was documented  in the town
of  Yauco (Puerto  Rico Department of Health, 1987).  The cause of the
outbreak was linked  to  the drinking water supply from the water
purification plant  at Ranchera  Ward, and, in fact, once  the plant  was
cleaned  and water  flow  returned  to  normal,  the outbreak subsided
(Puero  Rico Department  of Health,  1987; EPA, 1987).   Deteriorating
water  quality may be the source  of the  marked  increase in  cases  of
gastroenteritis,  dysentery,  and other undiagnosed diseases  being
reported by  The Puerto Rico Department of Health (from  29,455  cases in
1984 to  approximately 73,722 in   1988).
      Many  common bacteria of water, such as Klebsiella pneumoniae.
Aeromonas hvdrophila. Legionella pneumophila. Pseudomonas spp.,
Vibrio cholerae. Vibrio vulnificus. and Vibrio parahaemoiyticus. as well
as  their  indicator, Escherichia coli. are potential  pathogens of man
(Ha-zen  et al., 1987).  Many of these  organisms can be  isolated from
polluted  and unpolluted waters  in  Puerto Rico, although not all  of them
have  been linked  to waterborae disease  occurrence (Hazen  et al.,  1987).
Almost  aJl  pathogens  are transmitted  to water by fecal contamination.
For  this reason,  bacteria  found exclusively  and universally  in feces are
assayed  as indicators  of  possible  pathogens  in  the water.  Coliform
bacteria  assays are the most  practical, since assaying for  the  presence of
pathogens themselves  is  costly, difficult, and  time consuming.  The

-------
pathogens may be  in  extremely  low  quantities  in  the  water (v,hich
males it  difficult to isolate and  identify  them) and still  be  infective.
      Coliform bacteria  have  been traditionally  used  as indicators  of
possible  fecal contamination,  since  they  were thought  to be present only
when feces  had been  deposited  in the water  (Hutchinson  and  Ridgway,
1977).  Conforms  are  subdivided into total and fecal  coliforms.  All  are
gram negative, non-spore  forming,  aerobic or facultatively  anaerobic
bacilli that ferment lactose  at 35°C,  producing gas (Sergey's Manual,
1974).   More than 50  bacterial  species have  positive coliform reactions
and can be  found  in the feces of warm blooded animals.   Fecal coliforms
also possess  a characteristic higher thermal tolerance;  that  is,  the  ability
to  ferment  lactose  at 44.5°C.  This  characteristic  allows the  isolation  of
the  group through  the  use of elevated temperature  tests.  IL. coli.
Citrobacter  freundii. Enterobacter cloacae, and Klebsiella pneumoniae
are  members of this group  (Bergey's  Manual,  1974).   Assays for fecal
coliforms usually  focus on Escherichia c,oU. first described  in 1855 by
Escherich.  £.. coli  is  found in high  densities in the feces of  warm  blooded
animals.  This  bacterium  is the  major component  of  both  the total and
fecal coliform group (Lavoie,  1983),  and is nearly  always  found, in  fecal
contaminated water.   However,  other coliforms  may  be found  even
when £.. colj is absent.   £.. coli is also the coliform  most affected by water
treatment  procedures (APHA,  1985).    Among the  most common
coliforms  found in  both raw  and treated water  supplies,  £.nte_robacte.i
species  are  the most prevalent.   Other genera of  common  occurrence,
are Citrobacter and Klebsiella.
      The  underlying  principles  governing  good  indicator bacteria  (as
stated bv  Bonde in 1977)  are:

-------
       1.   Indicator bacteria  must  be  present  whenever pathogens  are
         present.
       2.   Indicator bacteria  must be present only  when  the  pathogens
         are  present.
       3.   Indicator bacteria  must be significantly  more abundant  than
         pathogens.
       4.   Indicator bacteria  must resist both  the  environment  and
         desinfectants  better  than pathogens.
       5.   Indicator bacteria must grow easily  on  relatively simple  media.
       6.   Indicator bacteria  must yield  characteristic reactions  so  they
         can  be unambiguously  identified.
       7.   Indicator bacteria  must be randomly distributed in the  water.
       8.   Indicator bacteria  must grow independently of  other  organisms
         on  artificial  media.
       The  problem of using  temperate water  quality  standards  in  the
 tropics has long been a source of discussion.  In a  study conducted  by
 Feachem (1974)  on the  waters of New  Guinea,  fecal coliforms (FC) and
 fecal  streptococci (FS) levels  were greater  than  100 CFU/100 ml,
 rendering  all  sites  unacceptable  as drinking water  sources.   He
 concluded  these  sites  were grossly  contaminated  with feces.   Yet FC and
 FS densities were  lowest  in  areas with high human population.  Evison
 and  James (1975)  reviewed  literature  from various  tropical  areas and
 concluded  that  E.. coli densities  at these areas did  not correlate  to
 sources of fecal contamination.   Several  other  studies  have
demonstrated  that  supposedly universal indicators  of  fecal
contamination were inappropriate for  use as such  in the  tropics.  These
include studies  showing  no  correlation  between  the indicator  organisms

-------
and  the  presence of Sa 1 mone 1 La.  sp. in v,ater  (Wright, 1982a;  Thomson,
1981);  studies reporting  high coliform and  streptococci  densities  in  non-
contaminated streams (Fujioka  and  Shizumura,  1985);  studies  reporting
the  occurrence  of alarmingly high  coliform  densities upstream  from
sewage  contamination sites  (Oluwande et al., 1983); and  studies
confirming  the low  accuracy  of  standard  media  for  the isolation of £,.
coli.  from tropical source water (Lavoie,  1983).
      In  temperate waters,  £.. coli  fits  most of Bonde's principles.  The
assaying  methods for this  particular  bacterium  take advantage  of its
thermotolerance  as opposed to  the lack  of thermal  tolerance of the
environmental  bacterial flora to  isolate it in  growth  media.   In  the
tropics,  this could be invalid, since components of  the normal
environmental  flora  can  also be  thermotolerant. Enteric  organisms,
being adapted  to the high  temperatures of the  gut  of warm blooded
animals  (usually around  37°C),  can survive and  grow in tropical  waters
(Bigger,  1937;  Ragavachari  and  Iyer,  1939),  especially since these
waters  are  high  in nutrient  concentrations (Hazen et al,,  1982;  Hazen
and  Aranda, 1981).   Studies  have shown  that  IL ££>U can persist in
tropical  freshwaters  (Carrillo  et  al.,  1985; Biam<5n and Hazen, 1-983;
Perez-Rosas  and  Hazen, 1988; Hazen et al.,  1982).  In fact, E. coli  seems
to be a  part of the normal  aquatic  flora of the tropics (Ha^en et al., 1987;
Bermudez and  Hazen, 1988;  Fujioka  and Shizumura, 1985).   Few studies
have  been done  on  the  activity,  survival, and density of coliforms in
tropical environments (Hazen et al., 1982;  Biam6n  and Hazen,  1983;
L6pez-Torres, 1982;  Ortiz-Roque  and  Hazen,  1983;  Fuentes  et al.,  1983;
Carrillo et al.,  1985;  Pe*le  et al., 1981).  It has  also  been  shown that
many coliforms  can  subsist as  free living  saprophytes (Bergey's  Manual,

-------
                                                                             0
 1974),  and,  as  such, can  multiply  and  form slimes inside  water
 distnbution pipes.   The isolation of high densities of the  same group at
 one  site may be indicative of such growth  (APHA, 1985).
      The  findings by The U. S. Water Resources Council~and The U.  S.
 Geological  Survey (where  most sites sampled  by  both  violated the
 maximum  contaminant levels  for  conforms) resulted in the
 condemnation of source water in Puerto Rico.   Yet recent  studies
 (Carrillo et al., 1985; Hazen  and Aranda, 1981; Lopez-Torres et ah,  1987,
 Sandago-Mercado and  Hazen,  1987, Rivera et  al., 1988; Bermudez  and
 Hazen,  1988)  indicate  that even water  from pristine  sites  in  the
 Caribbean  National Forest exceed  the  maximum  contaminant levels  for
 fecal coliforms, yet less than 40%  of 300  fecal coliform isolates were
 actually £. coli (Santiago-Mercado  and  Hazen,  1987).   Similar  studies
 involving the same methods  in  temperate  climate  zones  (Pagel et  al.,
 1982) demonstrated that 90% of all fecal coliform isolates  were £.. coli.
 This reveals that  the  environmental flora in Puerto  Rico  has bacteria
 capable  of  producing  false positive reactions when assaying for  the
 presence of fecal  coliforms.  Comparisons  of four fecal  coliform assay
 methods in Canada and Puerto  Rico yielded interesting  results:  The
 specificity  (ability  of  the  medium  to restrict  growth  of  organisms other
 than the target  bacterium)  of  the  media in Puerto Rico  was significantly
lower (at least  by 20%) than that recorded by Canadian investigators
(Pagel  et al.,  1982).  It seems that, in  Puerto Rico, all methods have
significantly higher false  negative  errors,  even though  the  accuracy of
methods was the same  on  the  island as  it  was in  Canada.   Santiago-
Mercado and  Hazen  (1987) showed that high  densities of  thermophilic

-------
                                                                            1 1
and  mesophilic  background  flora in tropical  waters significantly  reduced
the levels of £., coli  on standard  coliform  media.
      In addition,  it has been shown that, in Puerto Rico, many
pathogens  can be found in the complete absence of £. coli.  These
include Candida albicans  (ValdeVCollazo et  al.,  1987), Klebsiella
pneumoniae  (L6pez et al., 1987), Legionella pneumophila (Ortiz-Roque
and  Hazen, 1987), Vibrio cholerae  (Perez-Rosas and Hazen,  1988),
Yersinia enterocolitica (Eh'as et  al., 1988), and Aeromonas hvdrophila
(Hazen et  al., 1981).  Under these  circumstances, results  from recent
studies on  the culturability  of certain  pathogens  become frightening.
Studies by Xu et al  (1982), Colwell  et  al (1985), and Roszak  et al  (1984)
have  shown  that E. coli. as well  as  pathogens such as Salmonella
enteriditis, and Vibrio cholerae  may survive and  remain  pathogenic in
the  environment,  but  are  unculturable  on  standard  media.   Furthermore,
the presence  of  enteric viruses in tropical  source  water has  also been
documented in the absence of E_.  coli (Berg  and  Metcalf,  1978; Toranzos
and Gerba,  1988; Keswick et al.,  1985;  Rose et al.,  1975;  Cabelli,  1983;
Hejkal  et  al.,  1982),  yet  viral assays  are  rarely performed.
      The  survival characteristics of £.. coli  in  the tropics  also affect its
performance  as  an  indicator organism  in  these  environments.  In  1937,
Bigger  was the first to report the growth  of coliforms  in  tropical  water.
Then,  in  1939.  Ragavachari  and Iyer demonstrated  the survival  of
coliforms  for  several  months  in  tropical river waters.   Recent  studies  in
Puerto Rico (Carrillo et al., 1985; L^pez-Torres  et al,, 1987;  ValdeV
Collazo et  al., 1987;  Hazen et al.,  1987)  have demonstrated  the survival
and proliferation  of £L coli  in tropical  rain  forest  streams.  Studies have
also  demonstrated  that  under the conditions prevailing  in a  tropical rain

-------
forest, Candida albicans (Valdes-Collazo  et al., 1987), Klebsiella
pneumonia?  (Lop-ez-Torres  et al., 1987), Legionella pneumophila (Ortiz-
Roque and Hazen, 1987), Vibrio cholerae  (Pe"rez-Rosas and  Hazen,  1988),
Yersinia enterocolitica (Elfas et al., 1988),  Aeromonas rfydrophila
(Hazen et al.,  1981), and Salmonella typhimurium (Jimenez et al,, In
Press) have  survival rates different  from those of their indicator, £.. coli.
Differences in  survival rates  between  £., coli  and pathogens in tropical
source  waters hamper  its ability  to  indicate  the presence of  pathogens
in these waters.  Escherichia coli always survives much longer in situ  in
tropical  waters compared to  temperate  waters.   The  increased survival
rates further stress  the fact  that  temperate  drinking  water  regulations
are  unapplicable  to  tropical waters (Hazen,  1988).
      Escherichia coli  has  been  found on  epyphytic vegetation 15  rn
above the ground  in a  Puerto Rican rain forest (Rivera et  al.,  1988;
Bermudez and  Hazen,  1988).   This points  to  the  possibility  of naturally
occurring  E_. coli  strains in  some tropical environments.   These
environmental  isolates  have  biochemical and  physiological
characteristics,  plasmid  profiles,  coliphage  susceptibilities,  and antibiotic
                                                               i
sensitivities that are  very similar to those of clinical E_. coli  isolates
(Rivera  et ah,  1988).   In fact, the environmental  isolates  were found to
have more than 85%  DNA homology and  identical mol% G+C with £.. c_oJi
B.   Naturally occurring £. coll could account  for the results  of studies
showing high coliform  densities  in  the  absence  of a fecal contamination
source.   This finding is yet  another reason  why,  in the tropics, E_. £olj
does  not fit  the underlying  assumptions  of a good  indicator organism.
      The  above mentioned  studies, plus  the  fact that  coliforms have
been  shown  to  be less resistant  to chlorination than some  human

-------
                                                                             1 3
 pathogens (Berg and  Metcalf,  1978),  indicate  that  the  criteria  used  for
 determining  the  bacterial  quality  of  drinking  water in  Puerto Rico  may
 be  incorrect;  that is,  fecal contamination may be detected when in fact,
 there is none.   With this  in mind,  it  is not surprising that we find it
 difficult  to meet the  legislated standard  for drinking water  as  stated  in
 the  Safe  Drinking  Water  Act  (Public  Law 93-523,  1974;  modified in
 1986 to meet changes proposed by  the U.S. E..P.A. in 1983).   Two
 possibilities  remain:  either to change the indicator system  and  / or the
 maximum contaminant  levels  used,  or  to  enumerate  pathogens  directly.
 Studies  suggest  that  fecal streptococci  (enterococci),  enumerated  with
 selective  media, may be  better indicators of  fecaJ contamination  than
 fecal coliforms  (Hazea,  unpublished  data).
      Enterocccci are  a subgroup of the fecal  streptococci.  The  group  is
 composed of  Streptococcus faecalis. S^ faecalis subsp. liquefaciens. $_L
 faecalis subsp.  zymogenes and £L faecium.  These  organisms are gram
 positive,  non-spore  formers that  have  a  spherical  shape  and are
 arranged  in  pairs or chains (Bergey's Manual,  1974).   They  have been
 used as  secondary indicators  of  water quaiity, but  the problem  with
 coliforms  has  renewed  interest  in  them  as  primary indicators  of water
 pollution  (Dutka and Kwan, 1978;  Cabelli, 1983).   Fecal  Streptococci  have
 not  been  observed to regrow  in  natural  waters (Evison and  James,
 1975).   They have  short  survival  times  outside  their  natural habitat,  so
 all  detected pollution  must be recent  (APHA, 1985).  Streptococcus
 faecalis is commonly found  in human feces (Bergey's  Manual,  1974).
 Other fecal streptococcus  species,  such as S. bovis, and S^e.quinui are
 found in  feces of warm  blooded  animals (Bergey's Manual,  1974).
Enterococci have been proposed  as indicators  of human  fecal

-------
contamination (Berg  and Metcalf,  1978; Cabelli, 1983; Cabelli et a].,
1983).   Fecal Streptococci densities correlate better  than  those  of E. coli
and  total coliforms for the  presence of enteric  viruses (Berg  and Metcalf,
1978;  Cabelli, 1983).
      Media used  to enumerate enterococci, such  as  KF  agar, contain
sodium  azide, which interferes with cytochrome  oxidase  in  the electron
transport chain  of aerobic  organisms, limiting  unwanted growth
(McFaddin, 1980).  This is  an important fact in the avoidance of over or
underestimation  of actual bacterial  densities.   It  has been  suggested
that  the densities  of fecal coliforms and enterococci can be  compared to
determine  the origin of contamination.   Ratios  of 4.0 or above  indicate
human  fecal  contamination, whereas ratios of 7.0 or  less indicate animal
fecal contamination (Geldreich. and  Kenner, 1969).  The  fecal coliforrn  /
enterococci ratio has been  criticized because  of  differences  observed  in
the  survival rates  of enterococci and  fecal coliforms and  because of
differences  found  among individuals with  differing  diets.  Hill  et al.
(1971)  found  that  individuals  with vegetarian  diets  exhibited   greater
densities of enterococci in their bacterial flora.   Later, it was  found that
in environmental water above  20°C, £_,. coli  survived  longer  than
enterococci (Evison and James, 1975).   These  findings  suggest  that  the
fecal coliforrn /  enterococci ratios  can vary, affecting our  perception of
the  actual  source of contamination   as  well as the extent of  pollution.
The  method has  been  used;  however,  to examine the quality of waters
while  trying to  determine whether  pollution was  due  to  human  fecal
contamination or  to  storm waters (Feachem, 1976;  Wheater  et  al.,  1979).
      This   study will examine the   presence and  densities of  total and
fecal coliforms,  as well as enterococci, as  indicators of fecal

-------
contamination  of  drinking  water and  different methods  of  enumeration
of these  bacteria  in  order  to  establish whether  they  are  adequate
indicators  of bacterial pollution in tropical  treated drinking water.
Samples will be  taken  from raw water at various  purification  plants  of
the PRASA system  (influent:  water coming  directly  from  the  natural
reservoirs)  as  well as finished  drinking water  processed   in the plant
(effluent).   The  study will  determine  if water treatment   affects the
levels; of coliforms  or enterococci present in finished drinking  water
and  whether  these  organisms are good indicators of recent fecal
contamination  of  tropical source waters.   The  objectives  of this  study
are:
      1.  To determine  which  indicators best  point to recent fecal
          contamination  of  tropical  drinking water.
      2.  To determine  which  organisms  are  affected by  treatment at
          potable  water  purification  plants  and  whether  treatment
          affects  the levels of  fecal coliforms.

-------
                     MATERIALS  AND METHODS

      Study Sites.   Four filtration plants operated  by  PRASA were
selected for the  study:
      Rancheras Water  System, located on  road 371, Km  13.0  in  the
town of Yauco was  selected for  this study  since  it was  the source of the
etiologicaJ  agent of a  gastroenteritis  outbreak that occurred during  the
summer of 1987.   This treatment plant,  with a  processing capacity  of
1,368,000  liters  of  water per day (GPD),  serves  a population  of 1,440
individuals.   The raw  water, obtained from Duey River, does not receive
aeration.   The pH is adjusted  through the  addition  of  lime (CaO).
Aluminum  sulfate (Al2(S04)3) is added for flocculation purposes.   Algae
are  controlled during periods of slow flow  by the addition  of  copper
sulfate.  Records of the addition  of these chemicals are not  available.
Raw water is  prechlorinated at the  water  inlet using  chlorine gas at a
rate of 0.08 kg/24  h.   The  water is then deposited in  a  sedimentation
tank for  1  hour  and then  filtered by gravity  through  a rapid sand  filter.
Post chlorination occurs at  the  pipe  leading to the storage tank at a rate
of  0.2  kg/24  h.   The plant  has storage facilities  for 57,000 liters of
finished water.   This plant  has never had problems with odor  or  color of
finished  water.
      The  Villalba Urban System, located  on  road 513,  krn  1.2, receives
water from four sources:   El Guineo Lake,  which  has sediments that
affect  the  raw  water quality  (particularly  the odor and taste);  Aceituna
Lake, which receives water from  El Guineo  La.ke  through a  concrete
channel; Jacaguas River, which is probably  the  main  source  of
contamination,  since  there  is  a sewage  plant  discharge  line  about  4

-------
miles  upstream from  the  water intake  line;  and El  Semil  Dam, providing
spring  water  coming  through  rock  and discharged into a concrete
channel that  is  not protected  against  grazing  livestock (another possible
contamination  source).  This  plant  was selected  for  the study  because of
its  location relative to  the  sewage processing plant.  Since  a  water intake
is located downstream  from a sewage  discharge  line,  the raw  water at
the  plant contains high levels  of fecal  contaminants. The  water is
sedlmented for one hour,  filtered  through  a rapid sand filter,  and
chlorinated by  adding gas until  an average concentration  of 2.5  mg/L is
achieved.   Finished water  is stored in  8 tanks  having a total capacity of
2,698,000  liters.
       Guaynabo Water  Treatment Plant, located  on road  873,  Barrio
Frailes Altos  in Guaynabo,  serves  the  town of  Caguas  and some  sectors
of Guaynabo.   Raw water  comes from  tanks  in Aguas Buenas,  which in
turn receive water from Cidra  Lake.   This plant  was selected because  of
the  contamination  levels at Cidra Lake due  to illegal sewage discharges
as well as  by the distance  that the water has  to  travel  before  it reaches
the  plant (approximately 15  miles).  Both  of these factors might  affect
the  raw water  quality.   Water processing  is carried as previously
described:  no aereadon  is necessary,  CaO and Al2(864)3 aje added,
sedimentation  occurs  over one hour,  and water  is filtered  through rapid
sand filter beds.   Chlorination  is  achieved  by the  addition  of  chlorine  gas
until an average final  concentration of  1.4 mg/L  is  achieved in the
water.    Finished water  is  stored in 4  uncovered  tanks with capacity  for
about  133,000  liters  each.
      La Plata System,  located on road  872,  km  5.2  in Toa  Alia, serves
various sectors  of  Toa  Aha  and Bavamon.   This  svstem receives  water

-------
 directly  from La  Plata  Dam,  which in turn  receives water from  La Plata
 River.   This river  receives sewage discharges  upstream from the dam
 and is also used for primary recreation (possible pollution  sources).
 Water is treated  with aluminum  sulfate  to  form floes,  sedimented for
 one hour,  pre-chlorinated  (at  0.1  kg/24 h),  filtered  through rapid sand
 filters, and  post-chlorinated using  gas until the chlorine  content  of the
 waters reaches  an average of  1 mg/L.  Water is also  aereated.   In
 addition  to  lime  and  aluminum  sulfate, polyphosphates (for the  control.
 of corrosion),  and fluoride  are added  to the water.  Finished water is
 stored in a  subterranean  tank  with  a  capacity of  19  million  liters.
       Sampling.   Water  was  grab  sampled  into sterile,  one-liter  Nalgene
 screw-cap bottles  at  the  water intake and  at the  finished water  primary
 distribution  line.   Samples  from the  finished water line were collected  in
 the  same type  of container with an appropiate  volume  of  a sterile 1%
 aqueous   solution  of sodium thiosulfate added  to  neutralize  free  chlorine.
 All  samples were kept according  to  sample custody procedures
 described in Standard  Methods for  the Examination  of  Water and
 Waste water  (APHA,1985)  and transported  to  the  laboratory  within  5  h
 of  collection.
      Water Quality.   Water  temperatures  were measured in situ using a
 standard,  centigrade  scale  thermometer.  Total  and free chlorine levels
 were also measured  in  situ   with a Hach chlorine  test kit employing  the
color comparator technique (Model CN-66, Hach Co.,  Loveland, Colorado).
The pH  was measured  upon collection using a portable  pH  meter.
Additional,  small  volume water  samples were  collected into sterile
Whirl-Pak  bags (Nasco, Ft. Wilkinson,  Wisconsin)  and transported to  the

-------
                                                                             1 9
laboratory  for  turbidity  analysis  using  a Model  16SOO  Turbidimeter
         •>             1       /        W
(Hach).
Bacterial  Analysis;
      Total Direct Counts.   Bacterial cell densities and the percentage of
those  active in  protein  synthesis were  determined  by acridine orange
direct  counts (AODC).   For  AODC,  an appropiate volume  of the sample
was  filtered through 47  mm diameter,  0.22 jam  pore size,  polycarbonate
membranes pre-stained  with 0.07%  irgalan black  (Nuclepore,
Pleasantville, California).  After filtration, 2 ml  of acridine orange "(0.1%,
aqueous solution) was  added and  left on  the  filter  for 2  minutes,
filtered, and the excess  washed off with  sterile  water.   Stained filters
were observed  using  the 100X objective  of an  epifluorescence
microscope (Model 2071, American  Optical Corp., Buffalo, New York).
Total  direct counts  were determined by  the  average of  the  number  of
fluorescing bacteria  in  ten  randomly  selected  fields on  the  membrane
multiplied  by 98177.77   (the number of fields on the filter).    Acridine
orange binds  to nucleic  acids  and fluoresces  red  or green under  the
epifluorescent microscope depending  on  the  amounts of  RNA and  DNA
present in  the  cells  (Hobbie  et al.,  1977).
      Percent  Activity.  The percentage  of  cells active in protein
synthesis were  calculated indirectly from  the total  acridine orange
direct  counts. The ratio  of  red fluorescent  cells (having  greater amounts
of RNA) to the  total  number of  fluorescent  cells as obtaine-d in the
acridine orange  direct  counts  represents  the  proportion  of  cells  actively
transcribing their genetic code in order  to  produce proteins  (Hobbie et
al.,  1977).

-------
                                                                            20
      Membrane  Filtration  Studies,  Levels of total and fecal coliforms,
 as  well  as fecal  streptococci were measured  by membrane  filtration
 studies  according  to Standard Methods  (APHAJ985).  Volumes of 1, 10,
 100,  and 1000  ml were  used in  each case.   All  presumptive growth was
 confirmed by inoculating at least 10% of the colonies  on each plate  Into
 appropiate  Standard Methods (APHA.1985) media  as  follows:
      Colonies  presenting the typical  green metallic sheen on m-Endo
 agar  (Difeo Laboratories, Detroit, MI) after 28h incubation at 37°C were
 confirmed  according to their  ability  to  grow and  ferment  lactose  with
 the production of gas within  24  h on Brilliant Green Lactose  Bile  (2%)
 broth and Lauryl  Tryptose  broth  at 37°C (APHA,1985).  Colonies  giving
 positive  reaction  on both media  were subcultured on  McConkey  agar  for
 24 h at  37°C to  verify  their identity as lactose  positive, gram-negative
 organisms (APHA,1985).  All media  were  purchased from  Difco
 Laboratories.
      Fecal coliforms present a typical blue color on m-FC agar (Difco)
 after  24  h  incubation  at 45°C, while  all  other colonies  are gray.   m-FC
 agar  is  prepared  according  to  the  formulation  published by  Geldreich et
 al (1969) and supplemented with  10 ml of 1% rosolic acid  in  0.2'N NaOH
 solution  per liter.   Presumptive  fecal coliforms were confirmed  on EC
 broth  (Difco), which contains  0.15% Bacto Bile Salts No. 3.  These bile
 salts  inhibit spore formers  and  fecal streptococci  from growing while
 enhancing the growth of  E_. coll at  45°C (Hajna and Perry, 1943).
 Organisms positive for the production of gas on  EC  broth at  45°C  after
 24  h  were  streaked on McConkey  Agar  (Difco)  and incubated at  the
 same  conditions to  isolate  lactose  positive, gram negative  colonies.
Approximately 10%  of these organisms  per site  per collection  were

-------
identified  using  the  API-20E system  for  the  identification  of
enterobacteriaceae and  other gram  negative  bacteria  (Analytab
Products, Inc., Plainview, New York).
      KF Streptococcus  agar (Difco)  was used to assay  for  the presence of
fecal streptococci.  This  medium contains sodium  azide, which  interferes
with  the electron transport  chain  of  aerobic  organisms  (Kenner  et  aL,
1961).   The  addition of  1  ml triphenyltetrazolium  chloride  1% per 100
ml of medium results in a red color  of  the colonies for  easy
identification.   Color is produced  as  enterococci reduce the  tetrazolium
to  an acid dye  (Difco  Laboratories,  1984).   Presumptive enterococci  after
incubation of KF Agar at 37°C for 48 h  were confirmed on Azide
Dextrose broth (Difco)  incubated  under  the  same conditions.   Azide
Dextrose broth  has  the  same selective agent as KF  Agar.  Turbidity of
the  medium after 48  h  was considered  as positive for  the  presence of
streptococci.   The broth  was then streaked  on  m-Enterococcus agar
(Difco),  which does not recover such  fecal streptococci as 5.. bovis and £.
equinus (thus selecting  streptococci of human fecal  origin).   This
medium  was  found to  be  100%  selective when  assaying heavily  polluted
waters (Difco  Laboratories,  1984).   Reddish colonies on this medium
after  48  hours at 37°C  were  considered  positive for  the presence  of  fecal
streptococci  in  the  samples.
      Injured  fecal coliforms were  assayed using m-T7  agar  (Difco),  a
new  selective  medium  that  allows for  the improved recuperation  of
these organisms  from drinking water.   In laboratory  studies, this
medium  was  able to recover 86  to 99% more  injured coliforms  than  m-
Endo agar (LeChevallier et  al.,  1983).   m-T7  Agar  recovered nearly
three  times more coliforms  than  did  m-Endo  agar from drinking water;

-------
 and less  than 0.5% of all  m-T7  Agar colonies gave false  negative
 reactions (LeChevallier  et  al.,  1983).   Additional selectivity  may be
 obtained  through  the  aseptical  addition of 0.1 jig of penicillin G per ml
 of  medium after  autoclaving (LeChevallier et al., 1983). ~ The identity  of
 the  typical yellow  m-T7  coliform  colonies obtained after  incubating for
 24  h at 45 °C was  confirmed as described previously  for  fecal coliforms
 detected  by  m-FC  agar.
      Most  Probable  Number.   The most probable  numbers  of total
 coliforms, fecal  coliforms  and  fecal streptococci were  determined  using
 the  five  tube method  for  0.1,  1.0,  and 10.0  ml  water  samples and
 appropiate selective media (Lactose broth for total  coliforms;  EC broth
 for  fecal coliforms; and Azide Dextrose broth for fecal  streptococci), as
 described  in  Standard Methods for the Examination of Water and
 Wastewater (APHA,1985).   All media  was purchased  from  Difco
 Laboratories,  Detroit,  Michigan.
      For total  coliform bacteria, five  tubes  of  sterile  lactose  broth were
 inoculated per dilution factor  (i.e.,  5 tubes  with  10  ml  water,  5  with  1
 ml,  5  with 0.1  ml).  The  procedure was performed for both  influent  and
 effluent.   The number of  tubes  from  each  dilution  showing  gas
 formation  at  37°C after 48 h  was  recorded  and  the number  of coliforms
 looked  up on a MPN  table in  Standard Methods for the Examination of
 Water  and Wastewater  (APHA,1985).  Positive  tubes  from  the  lowest
 dilution  were confirmed  in the same  manner  as colonies  from m-Endo
 Agai (Membrane  filtration  studies).  Fecal  coliform bacteria  MPN
determinations were performed  in   the same manner,  but the  tubes
were incubated  at  45°C.   Positive  tubes  from the  lowest  dilution were
confirmed  exactly like the  colonies from  m-FC  and m-T7  agais.

-------
      The  MPN  technique to assay for fecal streptococci was  also
performed  with five  tubes per  dilution factor and  at 37  °C.
Confirmation  was carried out as with  colonies  isolated  on KF agar.
      Quality  control.  All  media  were prepared according  to  package
directions.   Unused  m-Endo  and m-FC agar plates were discarded after  2
days.  Unused KF and  m-T7  Agar plates  were discarded after 30 days.
Unused  liquid media tubes were discarded after 90  days.   All  media
were kept  in  a dark  cold room at 4°C
      Sterility of all media  was checked  for each  batch by  incubating  a
sterile plate or tube  at 37° or 45°C, depending on the  media.   Specificity
of all  media was checked for each batch of medium by inoculating  one
tube or plate  with E_. coli B  (ATCC 23848) and another  with
Streptococcus faecalis  (CBSC  15-5600A)  and incubating under the
desired  experimental conditions.
      Media  Performances   The relative performances of the viable
count  methods were  calculated for each site.   Analyses  of  variance  were
performed  on the average recovery efficiencies of each method.   If  a
site  revealed  significant  differences,  a Student  Newman-Keuls  test  was
performed  on  the  data  to  determine   the nature of  the differences.
Results from  these  tests  allowed the   rating  of methods  according  to
their  performance.
      Specificity  and Selectivity of the Media:  Specificity  was evaluated
by  confirming 10%  of  all  positive (presumptive target)  and  negative
(presumptive  nontarget)  colonies from plates containing  10  to 100  CPU.
The  false  positive  error (FPE)  was defined as the number  of false
positive  target colonies divided by  the total  number  of presumptive
target  colonies.   False negative  error  was defined as the  number  of false

-------
negative  target colonies divided by  the  number of  verified  negati\e
target colonies  plus  false negative  target colonies.   All results were
expressed  as percentages.   The  procedure was  repeated for  both
untreated  influents  and chlorinated  effluents.
      The selectivity index  is both  a  measure of the inhibitory effect  of
background  on target bacteria  and a  measure of the capability of the
media to prevent background growth  (Pagel  et  ah,  1982;  Clark,  1980;
Ray and  Specks,  1973;  Franzblau et ah,  1984).  If low selectivity indexes
are  found,  background  bacteria  are inhibiting the  growth of  indicator
bacteria  (Clark,  1980).   This parameter was  calculated  by dividing  the
number  of  presumptive  target  colonies  by the  total  number  of isolates
from each  media.   All  results  were expressed as percentages.
      Data   Analysis:   Statistical analysis  will be performed  with
programs developed  for Apple Macintosh Plus and  Macintosh SE
computers.  Multiple  correlation  and  regression  analyses  to correlate
bacteriological  and  physico-chemical  data were performed  using
Statview  512+ software  from Abacus  Concepts.   Analyses of  variance
and  Takey's  tests  to determine  differences  between sites, bacterial
densities, and physico-chemical  parameters  were  performed  using TCH-
Stats-ANOVA  software  developed by Dr. Terry C. Hazen.  The  log  (X + l)
transformation was  used to  make  heteroscedastic data   more
homoscedastic.   This transformation  was  selected due  to the exponential
variability  of the  data.   Any  statistical probability  equal to  or less than
0.05  was considered  significant  (Zar,  1984).

-------
                                RESULTS

      Water  quality.   Averages of water quality  parameters at .each  site
 are  presented  on Table 1.   These  measurements  were  subjected  to
 analysis of variance (ANOVA).   If significant differences  were detected
 by site, a Tukey's  Test was performed  on the data to  determine  which
 sites were different.   Measurements of  the  total  and  free  chlorine
 content  of  influents were  not  significantly different  by site.
 Temperature,  pH,  and  turbidity, however, differed by  site  (Table  1,
 Appendix  I).   Influent  water  temperature was  equal  only  between
 Bayamon and  Guaynabo (Tukey's  Test: Q=4.23,  df=40,  P>0.05), while
 Villalba and  Yauco  had significantly  lower values.  Influent water pH
 was significantly  lower  at  Guaynabo  (Q=3.71, df=40,  P<0.05),  while
 equal  at  the  other  three  sites.   Influent  turbidity  was  significantly
 higher  at Yauco,  but equal  at  Bayam6n, Guaynabo and  Villalba (Q=8.54,
 df=40,  P<0.01).
      Effluent pH  did  not  differ significantly  by  site (Table  1, Appendix
 II).   Effiuent  temperature,  turbidity,  and  chlorine content  was
 significantly  different  by  site (Table  1,  Appendix  II).
      ANOVA  between influents  and  effluents by  site  revealed a
 significant difference in temperatures  at Yauco  (F=10.56,  df=l  and 20,
 P<0.01).  Significant differences were  also found in the total and  free
chlorine content (F=55.5,  132.2, 133.4, 231.3;  df=l  and  20; P«0.001;  and
F=40.9,  162.4, 229.3, 202.2; df=l and 20; P«0.001, respectively)
between influents  and  effluents  at  all sites.
      Bacteriological  methods.   Average  bacterial densities at  each site
are presented  on Table  2.   ANOVA of the viable  count methods used

-------
                                                                            26
 revealed  no significant differences between m-T7 agar  and m-KF  agar
 influent  results  by site (Table  2,  Appendix  III).   In  Bayamon,
 significantly lower densities of  total  coliforms, fecal  conforms,  and fecal
 streptococci were detected with MPN methods than  in any  of  the  three
 other sites (Q=7.4, 6.8, 5.5; df=40; P<0.001; see Figs.  2 and 6).  None of
 the  methods  used was significantly  different  by  site  when assessing
 influents (Appendix   IV).
      ANOVA between methods used to detect  the same group  of
 organisms  in influent waters  revealed that total  coliform  densities  were
 significantly different between  membrane  filtration  and  MPN  (F=4.28,
 df=l,86,  P<0.05;  see  Figs. 2-5).  Fecal coliform detection methods  were
 also different (Figs.  6-9).  While  no significant difference was  found
 between  m-FC  agar  and fecal  coliform  MPN, membrane filtration  with
 m-T7  agar  produced  significantly  lower results  (Q=6.05  and 5.31,
 df=129,  P<0.01).   None of the  other  comparable viable count procedures
 for  total coliforms and fecal  streptococci showed significant differences
 when  assaying  effluents.
      ANOYA of methods pooling  sites  (Appendix V) revealed,  in the
 case of  influent water, the variability of  results  from  the  same method
 when used  to assess   water from different sources.   Only m-T7  agar  and
 m-KF  agar  gave  consistent results  between  sources of  water.  In the
 case of  effluent water, similar  tests  revealed  that for  most methods,
 water quality from  the four  plants  was consistent (Appendix V).
 Notable  exceptions were MPN  for  both  total and fecal  coliforms; some
 plants  performed  better  than  others  in  coliform  removal.
      AODC and  percent cells active  in protein synthesis. Average AODC
and  percent activity are presented on  Table 3.   ANOVA  of  influent  AODC

-------
did not  exhibit significant  differences by  site  (F=2.41, df=3  and 41,
P>0.05;  Figure 11).  Percent activity of the influents was  similar only
between Bayam<5n  and Villalba, while  those of  Yauco and Guaynabo
were  significantly  lower, as  revealed by Tukey's test (Q=4.86 and  7.54;
df=40;  P<0.05 and 0.01,  respectively).  Effluent  AODC was significantly
lower  at Guaynabo (Q=4.62,  df=40, P<0.05),  while  similar at  the other
three  sites.   Effluent  bacterial  activities  were significantly  different
between all  sites (F= 10.53, df=3  and 40, P«0.001).  Further analysis
(Tukey's  test) revealed  significant  similarity  between  the effluent
activities of Bayamon  and  Villalba (Q=l,34, df=40,  P>0.05),  while  those
of Guaynabo and Yauco  (Q=1.21,  df=40, P>0.05) were also similar
between themselves, but lower than the  other  two.
      AODCs  were  also  used  as  measurements  of  bacterial  background
relative  to  the  indicator  organisms  assayed.   The higher the ratios
between the indicators and AODC,  the  lower  the amount  of  background
bacteria present in  the  sample.
      Among influents (Table  3), the  highest average ratio  was obtained
when  assaying total  coliforms  in Guaynabo.   In  fact, this  site's  influents
had the lowest overall level  of background bacteria  relative  to  the other
sites.    On  the other hand, Influents from  Bayamon contained the  highest
levels  of background bacteria.   In  the effluents, the highest  indicator /
AODC  ratios were  those  of MPN methods  at Guaynabo.   MPN
methodology  generally had higher  ratios   than  those of  membrane
filtration.   The lowest ratios were those of Bayamon's  effluents.
      Correlation of bacterial  parameters  and water  quallty,_Q_f_LQnjjgj]t$
and  effluents.  Two correlation matrices   were  performed to  determine
the  relationship between  viable  count methods,  AODC,  the  percentage  of

-------
                                                                             28
cells  active  in  protein  synthesis,  and water  quality parameters  of
combined  influents and  combined  effluents.   All correlations  mentioned
were  significant at the  P<0.05  level.
      In  the  influents,  all  viable  count  methods  were positively
correlated  with  each other  (Table  4).   In the effluents (Table 5), total
coliforms, as detected  by  membrane  filtration,  were positively
correlated with  fecal coliforms  detected  by  membrane  filtration  with  m-
FC agar, and total coliform MPN  correlated positively with fecal
streptococci  MPN.
      Correlation of bacterial  parameters  and  water quality  at  each site.
Two  correlation matrixes;  one for  influent, one  for  effluent,  were
performed by  site to  determine the   relationships  between  water quality
viable count  methods,  AODC and  percent  activity.   All correlations
mentioned were significant at the  P<0.05  level.
      At Bayamon, (Table  6) the  influent  total coliform  levels  (as
detected  by  membrane  filtration)  correlated  positively  to  all  the  other
viable count techniques  assessed, AODC, and  pH.  Fecal  coliforms
detected  by  membrane  filtration using m-FC  agar were  positively
correlated  to  results of fecal  coliforms from  m-T7  agar, total coliform,
fecal  coliform,  and  fecal streptococci MPN,  and pH.  Fecal coliforms
detected by  m-T7  agar  correlated  positively  with total coliform  and
fecal  streptococci  MPN;   There are-positive  correlations  between total
coliform MPN and  fecal  streptococci MPN,  fecal  coliform  MPN  and  fecal
streptococci MPN, and  AODC and percent  activity.  Temperature
correlated  positively with   total  chlorine  content,  and  negatively with
total  and fecal  coliforms  detected  by membrane  filtration, total coliform,
fecal  coliform, and  fecal  streptococci MPN, and AODC.

-------
                                                                             29
      In  the  Bayam6n effluents  (Table  7), fecal streptococci levels
determined  by membrane filtration were positively  correlated  with
fecal streptococci MPN,  total and free chlorine  levels,  and fecal  coliforms
detected  by  m-T7 agar.   Fecal  streptococci MPN were  positively
correlated to total  chlorine  levels.   Turbidity  was  positively  correlated
to  m-T7  agar fecal  coliforms, fecal streptococci (detected by  both
membrane filtration  and  MPN),  and free  chlorine content.   A  positive
correlation  also  existed  between total  and free chlorine levels.
      The Guaynabo influents (Table  7)  displayed  positive correlations
between  membrane  filtration total  coliforms  and   membrane  filtration
fecal coliforms (using m-FC  agar);  membrane  filtration fecal  coliforms
(using  m-FC agar)  and  m-T7  agar,  membrane filtration fecal
streptococci, total coliform MPN, and fecal coliform MPN; m-T7  agar and
both  fecal coliform  and fecal  streptococci  MPN;   total coliform  MPN  and
fecal streptococci  MPN;  and  fecal coliform MPN and both  fecal
streptococci  MPN and  percent   activity.   Influent temperature correlated
negatively with pH,  m-T7, and  all   MPN methods.   Turbidity was
positively correlated  to  membrane  filtration  total  coliforms,   membrane
filtration  fecal streptococci, and  AODC.   Percent  activity  correlated
negatively with pH.
      In  the  effluent from the  Guaynabo plant  (Table  9), membrane
filtration  total  coliforms  were positively correlated to AODC.  Also,  total
and  free  chlorine levels  were positively  correlated.
      The correlation matrix  for Villalba influent (Table  10)  reveals
positive correlations  between  total   coliforms  detected  by  membrane
filtration  and fecal  coliforms  detected  by  m-FC agar and  m-T7 agar,
fecal  streptococci detected by membrane  filtration,  fecal  coliform  MPN

-------
                                                                             30
and  fecal  streptococci  MPN.   Fecal  coliforms  detected  by  membrane
filtration  using both m-FC  agar  and m-T7  agar  were also  positively
correlated  with fecaJ coliform  and  fecal streptococci MPN.   Fecal coliform
MPN  were positively correlated  with fecal  streptococci  MPN, as  were
AODC and percent activity.   Temperature of influent water  was
positively  correlated  to  total  coliforms  detected  by membrane  filtration,
fecal  coliforms detected by membrane  filtration using  m-T7  agar,  and
fecal  streptococci  MPN.  Influent  pH  correlated negatively  with total
coliforms  detected  by membrane filtration.   In  the  Villalba effluents
(Table 11),  total  coliforms  detected  by  membrane filtration  correlated
positively  with fecal coliforms  detected  by membrane  filtration  using
m-FC  agar, as did  fecal  coliform MPN and  turbidity.  Effluent  total and
free chlorine   content were  also  positively correlated.   Effluent  water
temperature  was  negatively correlated  to pH.
       In the influent at Yauco, all  the  viable count  methods for all  of  the
bacterial  groups  assessed  were  positively correlated to one another
(Table 12).   Temperature  also correlated positively  to  all  methods
except membrane  filtration  for the  detection of  fecal coliforms  using  m-
T7 agar.   AODC correlated  negatively  with  the percentage  of cells  active
in protein synthesis.   Other negative  correlations existed  between
influent  pH and membrane  filtration fecal coliforms  using  m-FC  agar, as
well as with all MPN  methods (total coliforms, fecal coliforms,  and fecal
streptococci).
      The  correlation matrix for  the effluent at Yauco  (Table  13) showed
positive  correlation between total coliform  MPN  and total  coliform
membrane  filtration, fecal  coliform  membrane filtration on  both m-FC
agar and  m-T7 agar.   FecaJ  streptococci  MPN  densities  were  also

-------
                                                                             3 1
 positively  correlated to all  these  parameters.   Fe-cal streptococci
 membrane  filtration densities  were  positively  correlated  to total
 coliform  and fecal  coliform  (m-FC)  membrane filtration.   Membrane
 filtration  for  the detection of fecal coliforms using  m-T7  agar  were also
 positively  correlated with these two  parameters.   AODC  correlated
 positively  to  both  total coliform  and fecal  streptococci  MPN, while
 correlating  negatively  with  percent  activity.   Turbidity  correlated
 negatively with pH.   The total and free  chlorine content  of the effluent
 were  positively correlated.
       Media  Performance.   Media performance when  screening for
 target bacteria  in  the  influents  was  not  significantly different  at
 Guaynabo  and Yauco (Table 14).   At  Bayam6a,  membrane  filtration on
 m-Endo  agar was  able  to recover  higher bacterial  densities than any  of
 the  other methods.   This  would be expected, since  this group should  be
 the  most  abundant  of  the bacteria  being  assessed.   At  Villalba,  not only
 was m-Endo agar able to  recover  high  levels of coliforms,  but  MPN
 analyses performed  better than  fecal coliform  and  fecal  streptococci
 membrane  filtration  analyses.   Evaluation of  media performance
 relative to  the  levels of  nontarget bacteria reveal that, at  Bayamon, m-
 Endo  agar  performed better  than  any other method  (Table  15).   At
 Yauco, m-Endo,  m-KF, and  total coliform  MPN performed  higher  than
 the  rest  of the methods as  far  as nontarget  organism  recovery.
 Guaynabo  and Villalba showed no  differences  among  media.
      Selectivity and specificity  of the methods  when assaying influents
 are summarized in  Table  16,  If  we  accept  15%  as  the upper level
permissible  for  false positives,  like Pagel et  al (1982)  did  when
evaluating  the performance  of fecal coliform  membrane  filtration

-------
                                                                            3 2
methods in a  temperate  climate  zone,  none of the methods  is  acceptable,
no  matter  how well the media  performed.   Media selectivity for
influents ranged  from 29 to 66% (Table 16).   Selectivity  of  the media in
tropical  waters was therefore much  lower  than those reported  by  Pagel
et al (1982)  in  temperate  waters.
      Performances in the assessment  of  both target  and nontarget
bacteria  in chlorinated  effluents  did  not  vary among media  (Tables  14
and  15).   However, false  positive  and negative errors in effluents  were
higher than those  of  influents,  and thus, also unacceptable.   Selectivities
in chlorinated  effluents  ranged from 47 to 79% (Table 17).   While  still
unacceptable,   these selectivities   are higher than those of the same
media when  assessing  influents.
      Among  the species isolated and identified from influents,  £.. coli
was  the  most  frequent (42.7%; Table 18).   The media isolating  this
bacterium  most often was  m-T7 agar.   MPN methodology recovered  the
lowest percentage  of  E.. coli.  Other frequently isolated species  were
Enterobactcr  sp. (27%)  and Citrobacter freundii (21%).  Enterobacter  sp.
was  detected  most  frequently  by m-FC agar,  while £.. freundii  was
observed  more frequently  on m-T7  agar.
      The  species  most commonly isolated in  chlorinated effluents was
also  £. coli (61.1%; Table  19).   m-T7 agar  isolated this bacterium far
belter. (50%) than m-FC  agar (27.3%) or FC  MPN (22.7%).  The  next most
frequently  isolated  species  was PjLejajiQJQjQlLiJ. sp. (16.7%).  m-FC agar,
apart from detecting practically  all Pjeudomonas  sp. encountered
(94.4%),  also  detected £.. freundii and E.. cloacae..

-------
                                                                             33

                              DISCUSSION

      Bacterial densities  and media performance  are  affected  in  various
 ways by a great number of physical  and  chemical parameters that can
 act  by  themselves,  synergistically,  or in  antagonistic ways  (Badge  et al.,
 1982).   Our results confirm  this fact,  as can  be  observed in  the  multiple
 correlations between the  collected  data.  The bacterial counts  we  see  are
 the  effects  of multivaiiate events.   In some cases,  these cannot be
 explained  by   the available data; however,  those  that can  are  not  only
 interesting, but of  great  relevance  to  the  issue of  drinking water  quality
 in  the  tropics.
      Influent temperature seems  to  be  a  very  important factor
 affecting the  levels of  bacteria.   Specific  patterns  are  evident,
 particularly in results  for the  most probable  numbers (MPN)  of  the
 bacteria under study.    Almost  all  of the  viable  count  procedures
 performed on  the influents of  La  Plata and Los  Filtros filtration  plants
 (Bayamon and Guaynabo, respectively)  correlate  negatively  with
 temperature (Tables 6,  8, 10, and  12).  There were no  significant
 differences  between the  influent  temperatures at  these two  sites,  while
 those of  Villalba and  Yauco were significantly lower.   Influent bacterial
 levels  at  Bayamon,  as  measured by  MPN,  were  significantly  lower than
 those of the  three other  plants.   All  of this leads  us to  beleive that high
 influent temperature depresses  the amount of organisms  present  in  raw
 water,   particularly  at  Bayam<5n.   This agrees  with  results  obtained by
Badge  et  al (1982)  from raw  waters in New Delhi.   Badge  and his
colleagues demonstrated  that  above  a specific  maximum temperature
level, rapid microbial  growth could  not  occur due  to  unrepairable

-------
                                                                           34
damage  to  the  cytoplasmic contents.   This finding  is particularly
important since it  has  been demonstrated  that £.. coli can die  very
rapidly with  slight temperature increases, while  other bacteria,  such as
Fseudomonas aeruginosa and Citrobacter freundii, which are
opportunistic  pathogens,   (both were  found  in  influent  samples  from the
four  study  sites; See Table 18) survive and can  produce  false positives
in  culture media  (Thomson,  1981).
      Temperature  is also  positively correlated  with almost all of  the
viable count  methods performed  on the Villalba  and  Yauco influents,
whose  temperatures  were  well  below  those  encountered  at Bayamon
and Guaynabo.   We suggest  that this  phenomenon is  caused by  an
increased survival  rate  of  the  microorganisms during periods  of higher
water  temperature  at these sites.   Our suggestion is backed  by the work
of  Wright (1982a), who observed  and described  a similar  phenomenon
while  working on  the survival  of fecal microorganisms in the  waters of
Sierra Leone, as well as by studies done  at our laboratory (Carrillo et al.,
1985;  Biamon  and  Hazen,  1983;  Perez-Rosas  and  Hazen,   1988; Hazen et
al.,  1982; Hazen et al.,  1987;  Bermudez and Hazen,  1988; Lopez-Torres,
1982;  Ortiz-Rcque  and  Hazen,  1983; Fuentes  et  al.,  1983;  Camllo et al.,
1985;  Santiago-Merc ado  and  Hazen, 1987; Hazen,  1988).
      Another  factor significantly affecting the  microbial  content of
water  is  turbidity.   Among the  studied influents and effluents,  increases
in turbidity bring about  increasing  total coliform, fecal  coliform,  fecal
streptococci, and AODC  levels, particularly at Guaynabo.   Observations
made during  samplings  point  towards  rainfall  as  the  cause  of increased
influent  turbidity.   The  clay  particles  and organic  matter that make up
turbidity can  adsorb ions,  molecules,  and  biological agents,  thus

-------
                                                                             35
affecting transport  and release  of toxic  materials, bacteria  and  viruses
(National Academy of Sciences,  1977;  Hoff,  1978).
      Influents  from  Guaynabo  and Yauco were the  most  succeptible to
elevated  bacterial densities due to increased turbidity -caused by
rainfaJl.  The sources of these  influents  are  not protected  against
roaming  cattle  and other domestic  animals,  or  from  human  activities
such  as  laundering.   We beleive these increased bacterial  contents  are
caused  by  the leaching of fecal and other waste materials from  river
and lake banks into source  water by rainfall.  The FC/FS ratios at  these
sites  under  these  conditions,   point  to mixed  human-animal  fecal
contamination during  rainfall  events  in a pattern is  similar to that
reported  by Barrel and Rowland (1977)  in Western Africa.   However,
routine   use  of  FC/FS ratios  might be  unacceptable  under normal
conditions  due  the the prolonged survival  and  normal presence of fecai
coliforms in  tropical  source  water.   Sediment  disturbances caused  by
increased water  flow  is another factor contributing to  the elevated
bacterial  counts.   Riverbed and  lake sediments  usually  have   higher
bacterial  levels  than  water due  to  the  adsorption of cells to clay and
other organic and inorganic  particles (National  Academy of Sciences,
1977).   Benthic  bacteria  washed  away by heavy precipitation  may  also
contribute to the  elevated  counts, since,  in  the  tropics,  these  might
include  bacteria  capable of false positive reactions on  standard  media
(Hazen,  1988).
      A direct  correspondence  between  increasing turbidity  of  Influents
and  effluents was observed.   Increases in effluent  turbidity  (above  1.5
NTU) were  also related to increasing bacterial  counts  at  all sites,
although  these  were  statistically non-significant.   The  statistical

-------
                                                                             36
discrepancy must  be related  to  the  lack  of  data  points,  since  the  above
mentioned situation occurred  a  limited  number  of times at  each site.
However,  increases in  effluent turbidity  bring  about increased detection
rates of fecaJ  streptococci  relative  to  the  detection of coliforms.
Detection  of  bacteria  under  these circumstances is possible  due to the
protection against  chlorination that  turbidity can  provide.   As  stated
previously, clay,  organic, and inorganic  particles  can adsorb  bacterial
cells.   These particles  also  have a  tendency  to  bind  together,  forming
clumps.  Bacteria  inside these clumps cannot be  effectively affected  by
chlorine; instead,  they  remain  sublethally  and  reversibly  injured
(National Academy  of Sciences, 1977;  McFeters  el al.,  1986;  World
Health  Organization, 1987).   These injured bacteria  remain  viable  but
unculturable  by  standard methodology.   It  has  been demonstrated  that
the  majority  of  viable  bacteria in  chlorinated  drinking  water are
attached  to paniculate  matter (LeChevallier et  al., 1988; World Health
Organization,  1987).  Furthermore,  the fraction of  turbidity  composed
by organic material can form compounds with  the  free  chlorine content
of  the  water,  thus  inactivating  it and  permitting  the survival  of bacterial
cells  (National Academy  of  Sciences, 1977;  Goytia,  1978; Hoff,  1978).
This  can result in the  eventual recovery of bacteria  in  the  water
distribution systems  due  to  increased nutrient  availability  and
favorable temperatures  (McFeters et  al.,  1986).
      In the  case  of fecal streptococci, we  suggest not only  the
possibility of  bacteria  protected against chlorine  through turbidity,  but
survival rates higher than those of  coliforms in the  presence  of chlorine,
as has  been reported with Klebsiella sp. (Ptak  et  ah, 1973)  and some
enteropathogens (LeChevallier and  McFeters,  1985), since  streptococci

-------
                                                                              3 7
were  found in  greater  numbers  than coliforms  whenever  these
occurred.   Defense  mechanisms  against  chlorination have  recently  been
described  for  several  bacteria (LeChevallier et aL,  1988).   Further
research  is needed to fully understand  the  behavior of- fecal streptococci
in  chlorinated effluents.   Correlations  between  the  detected  densities of
fecal  streptococci, chlorine content, and  turbidity (Table 9)  confirm
these  possibilities.   Turbidity values were  a highly significant  factor
affecting both the chlorine content  and  fecal  streptococci  levels at
Bayamon.  At the same time, chlorine  had a highly significant effect on
fecal  streptococci  levels.  All of these correlations  point towards
turbidity as the  factor determining  the  occurrence  of fecal  streptococci
in  drinking  water through  interference with  chlorination.   Fecal
coliforms  were also  detected  in  Bayamon effluents by m-T7  agar, in a
pattern  similar to  that of  fecal  streptococci  in terms of the  relationship
between  bacterial  density,  turbidity, and  chlorine content.   The fact that
fecal coliforms detected at this  site existed  in an injured  state  (they
were detected only  by m-T7  agar)  is further evidence  of  the  protection
against  chlorination  that turbidity  particles  can  confer  to  bacteria.
       Wright  (1986)  stated that  the  flushing  produced  by heavy rainfall
resulted in periods of increased  health  hazards due to  the  introduction
of  pollutants  into  bodies of water used  as drinking  water  supplies.   Our
observations suggest  that  heavy  rains bring  about periods of  highly
uncertain water quality.  If E^ QoJi is  naturally occurring in  the  tropics,  it
would  be  difficult to  determine  whether increases  in its  density during
heavy  rainfall  axe  due to the flushing of strains  from fecal  material  or  to
the flushing of  strains from  the  environment  into  the reservoirs.   Heavy
rains  also flush  soil   surrounding  the  reservoirs  and disturb sediments.

-------
                                                                              38
The resulting  high turbidity  affects  the efficiency  of  chlorination and
supports  high  levels  of  background  bacteria  (which interfere with
coliform  recovery on standard media).   High  turbidity  also clogs  filter
membranes  and  makes  it difficult to  properly  observe "microbial growth
in  fermentation tubes.   Special  care  should be  taken  when  analyzing
potable  water  samples  coming from  high  turbidity sources,  rainy  season
or  not;  particularly if  their  turbidity  is even  slightly  higher  than normal.
We also note  that while bacterial counts  are  increased in both  influents
and effluents  after the  onset of  rains,  so  are  the numbers of false
positives and  negatives  found.   We agree  with  Wright (1986)  that
membrane  filtration is  the  method of  choice  under these circumstances,
since  it provides relatively pure  colonies that  aid  in the  process  of
confirmatory subculturing in appropiate media.   However, since high
turbidity  can clog  filters, special  care  should  be taken in the preparation
of  suitable  dilutions to overcome this problem.  We go further,  to
suggest  an   actual identification  of a  suitable  amount  of  nontarget
organisms from each fecai coliform negative  sample to be absolutely
sure of  the  absence of  fecal  contamination  of  potable  water.   It  is
important  to remember  that  while high turbidity  will  not necessarily  be
an  indicator of potential  health  hazards, low turbidity   is  not an
inequivocal  indicator  of  the  bacterial quality  and  safety  of  potable
water.
       All of the  methods  used to  monitor  the presence of indicator
organisms in tropical source waters  seem  to be  unacceptable  due to
their high rates of false  positive and negative errors (Tables 16  and  17).
Tropical  waters have  autochtonous bacteria capable of producing
positive  coliform  reactions  when  standard methodology  is used

-------
                                                                             39
(Santiago-Mercado and  Hazen,  1987;  Hazen,  1988).   These bacteria  ca-jse
high  numbers  of false  positives on  standard  media,  which  results in
overestimatioo of pollution.   The  acceptable  limit for false negatives
used  by Pagel et al  (1982) in temperate waters was  5%~.   In  tropical
waters, we found false  negatives  ranging  from  12  to  50%  (Table 16).
Thus,  according  to  specificity indexes,  all the  methods tested  are  grossly
unacceptable  in  tropical  source  waters.   In addition,  the high  levels  of
background  bacteria  present  in tropical waters are capable  of producing
metabolic  interferences  with  the  normal  reactions of indicator
organisms  on  standard media (Means and  Olson,  1981; Burlingame et  al.,
1984;  Santiago-Mercado  and Hazen,  1987;  Hazen,  1988).   This
interference  can  result  in  underestimation  of  the  levels of  pollution in
tropical  waters.   The levels  of background bacteria relative to each  of
the indicators measured can  be determined  indirectly from the ratio
between the indicator and  the total  direct  count (determined by  AODC).
These  ratios range from 0  to  1.08  x 10~4.   The indicator bacteria being
assessed by  standard media are only 0  to  0.01%  of  the total  microbial
flora  present  in  the  analyzed  waters (Table 3).  The  low  selectivity
indexes  recorded  (29 to 66%  for  influents, 47  to 75%  for effluents;  see
Tables  16  and 17) relative  to  the  performance of  fecal  coliform
assessment methods  in  temperate  zones (Pagel et al.,  1982)  are further
evidence of the  inhibitory  nature  of  background  flora (Clark,  1980).
      Final identification of  colonies  from fecal coliform methods  are  yet
another  measure  of  the  high density and  variability  of background
bacteria.   Among influents,  £.. coli  was  successfully isolated 42.7% of  the
time out of  170 isolates (Table 18).  Other common  nontarget bacteria in
influents were Enterobacter  sp. (15.8%) and Citrobacter freundii  (12.3%)".

-------
                                                                           40
The Enierobacter and  Citrobacter genera have  positive  fecal  coliform
reactions, but are not fecal specific  (Bonde,  1977;  Hoadley et  ah,  1975;
Santiago-Mercado,  1987).     Identification  of organisms  recovered from
chlorinated  effluents by  fecal coliform  methods resulted" in the  isolation
of £_. c_pJi 61% of the times out of a total of 43 isolates (Table  19).  Other
bacteria frequently  isolated  from  effluents  were Pseudomonas sp.
(16.7%) and Enterobacter cloacae (13.9%).  The occurrence
of Pseudomonas sp.  in chlorinated effluents is important, since  this
bacterium, if ingested  by humans, might cause various  sorts  of
infections.
      Confirmation  rates of  both  target and  nontaiget bacteria (Tables
18  and 19) reveal  that a maximum  of 61% of all  colonies were actually
£.. coll by any fecal coliform  method.  This  is lower  than results recorded
by  Santiago-Mercado  and Hazen  (1987), who found  that  a  maximum  of
70%  of tested  colonies were  E_. colj in  various types  of tropical
freshwater.   By  comparison,  Pagel  et al (1982) recorded confirmation
rates  that  were  at  least  27% higher in contaminated temperate
freshwater.   Pagel  et al (1982) found high numbers of E_.  coii   among  the
nontarget colonies  assayed.   She  attributed  her  results  to the thermal
stress  imposed on  the indicator by the colder  waters of temperate
climates.
      Comparisons  of the densities  of  target  fecal coliforms recovered
from  influents  reveal  that m-FC  agai had higher  recovery  rates  at
Bayam6n  and Guaynabo (Table 14).  At Villalba and Yauco, however,
fecal  coliform  MPN  had  better recovery rates, as  would   be expected
from  highly  polluted waters  (Santiago-Mercado  and  Hazen,  1987).   The
high bacterial densities  in  heavily polluted waters mask  fecal coliforms

-------
                                                                             4 1
by  inhibiting the expression  of their  normal   characteristics.  Densities
of  total  coliforms  were higher at  all  sites  when  m-Endo agar was used
(Table  14).   The highest  overall  total coliform recovery was recorded  on
the VillaJba influent,  which is grossly polluted  by a sewage outfall.
      Influent fecal  streptococci  were recovered  most  efficiently  by m-
KF at Bayam<5n  and Guaynabo (Table  14).   At Villalba; however, the
fecal  streptococci MPN method performed  better.   This could also be
explained  by the better overall performance of MPN  methodology in
grossly contaminated  waters  (Santiago-Mercado and  Hazen,  1987).   It  is
interesting  that at Yauco,  the   membrane  filtration and  MPN methods
had  similar  performances.
      Recovered  bacterial  densities in  chlorinated drinking  waters  by
the methods  tested reveal  that MPN    methods generally had better
performances (Tables 14 and  15).   This fact had  been noted previously
by  Grabow  et al (1981) in temperate  waters.   Notable exceptions were
m-KF recovered  fecal  streptococci  at  Bayam6n  and Yauco.   At these two
sites, membrane  filtration  was able to detect  fecal  streptococci when
MPN failed to do so.  This could  represent the  survival of  fecal
streptococci  to  standard  treatment.
      Influent nontarget  bacteria  recovered on  fecal  coliform  media
were  lowest when m-T7  agar was  used (Table  15).  The only  site where
this was not  true was  at Guaynabo.   The water source  at Guaynabo is
Cidra Lake, which has been  said  to be grossly polluted due to  illegal
domestic sewage  discharges.   Since the lake is very distant from the
treatment facility,  water  has  to  travel  inside pipes  for  approximately  15
miles  before  reaching the  plant.   It is possible that  the  high pollution
level  of  the lake, combined with  a possibly  thick  biofilm inside  the

-------
                                                                            42
 pipes, provide  optimum  conditions  for  the  growth of bacteria  not
 normally associated  with water  sources.   These  unusual  bacteria might
 be  the-cause of the higher nontarget levels detected by  m-T7  agar  at
 Guaynabo,   This fact  is supported by  the  observation that most of the
 unidentifiable  bacteria reported  in Table  18 came from  Guaynabo
 influents.
       Apart  from  the  low  levels of influent nontarget  bacteria reported
 on  m-T7 agar, this  medium  also recorded the highest selectivity  index
 among the  fecal coliform methods tested (Table 16).   The specificity  of
 m-T7  agar  was also the  closest to the  limits adopted by  Pagel  et  al
 (1982) for  temperate waters.   Perhaps  the  low performance of this
 medium when assessing  nontarget colonies (Tables 15 and  16) is due to
 the ability  of  m-T7  agar to  suppress some of  the background  growth
 associated with tropicaJ  source waters.   m-T7 agar also  recorded  low
 levels of confirmed  nontarget  colonies  (Table 16).  This  might  be an
 indication that  m-T7 agar,  while still  unacceptable for fecal pollution
 assessment  in  tropical  source  water,  might be  useful for  fecal  coliform
 isolation  in  this environment  with a better chance of inequivocal
 recovery  in  one step.
       Studies have  shown  the influence of  the type  of  water analyzed,
 the media used, and techniques   employed on  the bacterial levels
 detected at  particular sites (Evans et al.,  1981b).   We found that
 although  the  levels  of total coliforms  detected  on  the influents by
 membrane filtration  (MF)  and MPN techniques were not different at
 Villalba  (Figure  5),  they did  differ  at  the  other study sites  (Figures  3-5).
A  similar pattern was  evident with  methods for  the  detection  of fecal
coliforms,  Bayamon  displayed no differences  between  methods for  fecal

-------
                                                                              43
coliform  assessment  (Figure  6),  while  at   the  other  three  sites,  the
methods  disagreed on the results  (Figures 7-9).  Fecal  streptococci
assays  revealed  differences between the MF and MPN-methods only  in
the  Villalba effluent  (Figure  10),  the  other  three  sites having  similar
levels detected by each  of the two methods used  in both influents  and
effluents.   All influents  had  high  total  bacterial densities, as determined
by  AODC (Figure 14).
      In  the  sites exhibiting  differences in the  influent  total coliform
levels detected by MF or  MPN, MPN  values  were slightly lower  than
those obtained by MF.  This  was  caused  by  non-coliform interference  on
the  culture media used.   The  MPN  technique has been criticized  for its
failure  to detect total coliforms  under particular circumstances  (Evans et
al.,  1981a; Evans et  al 1981b;  Jacobs  et al., 1986).  Evans demonstrated
that if  the total bacterial  densities exceed  500  cells per  ml,  the  MPN
technique results in  decreased coliform densities  being  reported due  to
competition  of non-coliforms  for nutrients, as  well as by the production
by  these  of coliform  inhibitory substances.  The influents studied had
high  total bacterial densities,  plus  high percentages  of  false positives
and negatives  resulted  from  their  assays,  the  lower densities of
coliforms  in source waters as  detected by  MPN  must be  a result of
inhibition  and masking by non-coliform organisms.   The  same
explanation  applies  to differences  between the  MF and  MPN methods
used  in  the  detection  of  fecal coliforms in tropical source  water.
      At  Villaiba and Yauco,  membrane filtration  using  m-T7  agar
resulted  in significantly lower  densities than  MPN  or MF  with  m-FC
agar.  As already discussed,  high  numbers of  false  positives and
negatives   occur  when  assaying  for  the  presence of fecaJ coliforms  in

-------
                                                                             44
tropical  source waters (Sannago-Mercado  and  Hazen, 1987).  Examining
the  confirmation  rates we  obtained for our fecal coliform  assay
methods, we  agree  with  these  results.   We also found that  while  the
fecal coliform densities detected by  MF with  m-T7  agar were lower than
those detected by m-FC  agar,  this medium has a higher reliability of
actually  detecting  the  target  fecal  coliform  rEscherichia coli) than other
methods.  MF with m-T7  agar produced  the  highest confirmation
percentage of any of the  three  methods used.   This  is  in agreement  with
results  obtained  by LeChevallier  at  al (1983),  who  stated  that m-T7
agar not only recovered fecal coliforms from chlorinated water,  but
could  also  be used to detect coliforms in untreated  surface  waters with
a  high  reliability.   MPN,  on the  other hand,  produced  the highest
percentage  of unidentifiable  microorganisms  plus low  confirmation
rates, confirming its  unreliability  due  to  common components of tropical
bacterial  flora being able  to  grow  in the media.
      Examination of the  species  isolated  from source  waters by fecal
coliform  assays (Table 18) reveals that m-T7  agar was able  to detect E..
coli 4.2.7% of the time in  the influents, as well as other organisms of
clinical  importance, such as  Shigella dysenteriae  and Klebsiella-
p.ne.u maniac.   All  of  this  suggests that m-T7  agar may be more  accurate
in the detection  of the target fecal coliforms  in tropical waters.   This
medium   performed equally  well  on chlorinated  drinking water.
      The occurrence  of fecal  coliforms in finished  drinking  water was
rare and most frequently  related   to increases  in  effluent  turbidity.
However, there were  a few  cases  when m-T7  agar  was able  to detect
fecal coliforms in the  chlorinated  effluents when  m-FC  agar and  MPN
could not.  In cases when  E_. coli  was detected in the effluents by  MF

-------
 using  m-FC agar, m-T7 agar was able  to  detect 45% more  £_. coli than rn-
 FC  and 55% more  than MPN.   This confirms the performance of m-T7
 agar in tropical source waters in levels similar  to  those stated  by
 LeChevallier (1983) for temperate climates.  However, since £,. coli
 might  be  a normal inhabitant  of tropical source water, the "accuracy" of
 m-T7  agar may be useless in  the tropics.  Since pathogenic bacteria such
 as Klebsiella pneumoniae. Shigella dysenteriae. and Pseudomonas  sp.
 isolated from  m-T7 agar  were isolated in higher proportions than  in  the
 other  fecal  coliform methods (Tables  18  and 19),  colonies  from m-T7
 could  be  screened  for the  presence of certain pathogens  in tropical
 source water.   This approach  towards  the  direct assessment of
 pathogens could save  time and  effort,  so its possibilities  should be
 further  evaluated.
       Results from the fecal streptococci assays on  effluents  also  suggest
 them  as  adequate  indicators of  fecal  contamination  of waters.   We have
 already  discussed  the  occurrence of fecal streptococci in  source  waters
 and  the relationship between increases in their  levels  with rainfall.  At
 first,  the  occurrence of  fecal  streptococci in chlorinated drinking  water
 seemed  to  be  strictly  correlated  to  the occurrence  of coliforms,  but in
 later samplings, fecal  streptococci  were  detected in the effluents  from
 Guaynabo and  Villalba even in the absence  of coliforms.   These
 occurrences  could  not  be  correlated to  any  of  the parameters  under
 study.    Three possibilities  exist: either  the  fecal streptococci detected
 were able to withstand chlorination and  recover from injury on the
complex  media used,  fecal coliforms  were so  injured  that  they  failed  to
recover even on m-T7  agar; or,  as  is the case  with coliforms,
components  of  the  normal  microbiological flora  at  the  study  sites  can

-------
                                                                             46
produce false  positives  on these  media.  Since  all media  used to isolate
and confirm  the  presence  of fecal streptococci  contain sodium  azide,  any
organism capable  of producing false  positives  on these media  must not
rely  on an electron transport  chain  with cytochrome  oxidase  for  energy,
or  they must  be  resistant  to  this chemical.  The high  selectivity indexes
of  FS  methods relative  to other assays studied confirm this fact
However,  studies  in Puerto  Rico  have demonstrated  the prolonged
survival rate  of Streptococcus faecalis.  the  target organism, in a tropical
rain forest stream  (Muniz et  al.,  In Press).   High  densities of S_. faecalis
were also found  to survive  in tropical  marine  waters contaminated with
petroleum  (Muniz  et al., In  Press).   All  of these  findings discard the
possibility of fecal  streptococci  as  adequate indicators of tropical
drinking  water  quality.
      The  issue  of  injured bacteria in tropical  drinking water quality  is
very  important.   Injured bacteria  have been known  to recover  in
distribution  systems due  to  favorable temperatures  and  increased
nutrient availability, posing  health  hazards  to  the community  (McFeters
et  al.,  1986;  Wright, 1932b).  In the four filtration  plants  studied, the
total  bacterial densities, determined by AODC  were  equal  among
influents and  effluents  (Figures  12-15).   This  suggests that  chlorination
might only be injuring  coliforms,  thus disabling  them to  grow  on culture
media,  but it is am effective in  the  disinfection of water.   Several
mechanisms  for  the survival of bacteria  in  chlorinated water have  been
proposed.   Attachment  or association  of bacteria with various  surfaces
can  promote  increased  resistance  to  chlorine disinfection  (Levy  et  al.,
1984; Tracy  et al.,  1966; Silverman et al.,  1983;  Camper et al.,  1986;
Hejkal  et al  1979; Hoff, 1978; LeChevallier et al., 1981).   The  production

-------
                                                                             47
of  extracellular capsules can  also  protect  some  bacteria against  chlorine
(Reilly  and Kippen,  1983;  Clark,  1984).   Members of the natural
microbiological flora  of  tropical  waters may be  capable  of  surviving  the
standard  disinfection  procedures for  the  production  of "potable  water.
Effluent bacterial activity was lower than  that of influents at all  sites;
however, at  Guaynabo,  Villalba,  and Yauco  activities  were not
significally different  between  influents and effluents  (Figures  16-18),
providing  further evidence  of  the  little impact  of the  disinfection
procedures on the aquatic  microbial  communities  at these sites.   This is
startling, since  high  levels  of total and free chlorine  residuals (ranging
from 1.68 to  2.45 mg/L;  See Table  1) were recorded  in  the  effluents  at
all  sites.   The concentration at Yauco  was high enough  to eliminate
Giardia sp. cysts (Rodriguez  et ah, unpublished data).  Since  these cysts
are  resistant  to  regular  chlorination,  the  possibility  of overchlorination
cannot  be  ruled out.   Neither  Giardia sp. nor Crvptospondium sp. cysts
were found at any of the other sites.  However, problems along  the
distribution line,  such as  broken pipes and collapses of  bacterial  biofilm
inside pipes   may be  hampering  the  performance of residual  chlorine,
thus  allowing  high  densities  of bacteria  to repopulate  the finished
water.   Overchlorination can  result in yet  another problem:  the
formation  of  trihalomethanes.  The  occurrence  of trihalomethanes is  due
to  the  oxidation  reactions  in  water  treatment  (chlorine  disinfection)
with  organic  substances  present  in the water ("World Health
Organization,  1987).   The  most  common  trihalomethane, chloroform
(CHC13), produces cancer in  laboratory animals (World  Health
Organization,  1987).   The  carcinogenic potential of other common
trihalomethanes, such  as  CHBrCl;,  CHBr:Cl, and CHBr}  (formed in the

-------
                                                                             43
casual presence  of  the  bromide ion) is still  being tested (World Health
Organization,  1987).   The formation  rate  of trihalomethanes  is
accelerated by  high temperatures  and  pH  as long as free  residual
chlorine is still  present  in the  water.   Total trihalomethane
concentrations in drinking water can go  up to 1000  ng/L  (World  Health
Organization,  1987).
      Effluent activity at Bayamo'n  was  slightly  lower than that of the
influents  (Figure 19).   This  could be  due  to a slightly  better performance
of  the  treatment process at  this site.   However,  this plant  also  adds
fluoride to effluents as  part  of the treatment process, and  the  impact  of
this chemical  on the microorganisms  studied in  this  research was not
measured.   This is  an area needing further investigation.   We also note
that the percentage  of  bacteria active in protein synthesis in  any sample
was highly variable  at each  site,  probably  due to the complex effects  of
a  high  number  of  physico-chemical  and bacterial parameters  and  the
interactions between them.   Another  possibility  may be the  seeding  of
coliform bacteria in finished water  by the sand  filter beds  used in the
treatment  process.    McFeters et  al (1986)  have   shown  that in temperate
waters,  up  to  6,300  coliforms can be   associated  with each  grain of sand
in a filter bed.  These  coliforms  can  actually be  inoculated into finished
drinking water,  as  water pressure dislodges clumps  from  the filter bed
that can survive chlorinadon due to  their  close  association with  each
other (McFeters  et  al.,1986).  The higher densities of both  coliforms  and
false  positive  coliforms  in tropical  waters might accumulate  in  much
higher numbers  in  the filter  beds, especially if  these are not
backwashed every day.   Thus,  in the  tropics,  coliforms  seeded into

-------
                                                                             49
 drinking  water may be  a common occurrence.   This adds  further
 uncertainty  to  the  problem  of  drinking  water quality  in  the tropics.
       These findings point towards  the  need for closer  and careful
 monitoring  of  the  methodology currently being  used  to" determine
 drinking  water quality in  the tropics.  We  also  see  the need for  the
 identification  and  proper management of water resources  of relatively
 variable  and reliable high quality  to  be  used as potable  water sources;
 this conclusion  has been  supported  by  other investigations  of  both
 tropical  and temperate  drinking water quality  (National  Academy  of
 Sciences,  1977).  Further research  on  the effects  of  physical  and  bacterial
 parameters and  their interactions  on  water  quality  would  also be
 needed at each  of  these water  sources to develop and  manage  them  in
 the  most effective  way.    A  better indicator  of  water quality is  needed to
 solve  the problems of  water pollution control  and  drinking water
 quality in the  tropics.   Direct enumeration  of pathogens,  while  costly,
 could  be a  solution.   However,  other indicators have been  proposed
 (Dutka,   1978).
       Bifidobacterium   sp. have been proposed  as tropical  water  quality
 indicators  (Evison  and James, 1975).   Since  B ifidobacterium sp.-are
 obligate anaerobes, they  are  not  able to survive in  most  extraenteral
 environments.   In  addition,  Bifldobacteria are commonly found in  the
 gut  of humans, but not  in animals  (Hazen, 1988).  Carrillo  et al (1985)
 showed that  the densities  of Bifidobacterium adolescentis.  decreased one
 order  of  magnitude per  day  in  a  tropical rain forest stream in  Puerto
 Rico.   These investigators also found that  the  densities  of  bifidobacteria-
ILke  organisms were greater  than those of £.. colj at  all sites, except  those
at  a  sewage  outfall.  The  media  developed   for the isolation of

-------
                                                                             50
bifidobacteria  (YN-6)  had  low specificity  if  background  anaerobic
bacterial  densities were high  (Carrillo et al., 1985).  In fact, less  than
80%  of isolates  from  a Puerto Rican rain  forest stream were actually
bifidobacteria.   Bifidobacteria shows  promise as an  indicator of  human
fecal contamination  in  terms  of their specificity  and  lack  of survival  in
situ.  The development of media  with  better specificity  in  tropical
waters  would  make  bifidobacteria excellent  indicators.
      While Clostridium perfringens. another anaerobe,   has been
proposed  as  yet another indicator alternative  (Fujioka and. Shizumura,
1985;  Bisson  and Cabelli,  1980), its densities in human feces from  the
tropics  are much lower than  those of  the  currently  used indicators
(Wright,  1982a).  In addition,  clostridial spores  can  resist disinfection if
the  chlorine concentrations and  pH  are inadequate.  These  facts  propose
serious  doubts  as to the ability to  detect £.. perfringens easily when an
eminent danger of pathogens  exists  (Wright,  1982a;  Hazen, 1988).
Wright(1982a)  could  not  correlate  the  isolation  of  £.. perfringens with
the occurrence  of Salmonella sp.   in tropical source water.   Furthermore,
Carrillo  et al  (1985) found  that the densities of total  anaerobes  were
high  in  sites  that had received torrential rainfall.   This, along  with the
increased  densities  of £.- perfringens  reported  in  uncontaminated  sites
after  rainfall  (Fujioka  and  Shizumura,  1985),  suggests  that  £..
pej fringes might  have an extraemeral source  in  the  tropics (Hazen,
1988).   While  bifidobacteria  and Clostridium perfringens might be  more
suitable  indicators for the  tropics,  we  must also take  into  account  that
techniques  involving the assessment  of these  anaerobes  require more
sophisticated  and  expensive techniques  (Franzblau et  al.,  1984).

-------
                                                                             5 1
      Recent  studies in Puerto Rico  have posed  coliphages  as  more
 specific  than  £.. coli in terms of pollution detection.  Toranzos  and Gerba
 (1988)  demonstrated  that  coliphages were  only detected in  water
 receiving sewage  effluents.   The  viradel  technique (Goyai and Gerba,
 1983) was  used to concentrate  the viruses  in one liter  water  samples.
 Escherichia coli C3000 (ATCC  15597) was  then  used  for  coliphage
 assessment.  Since the assay  is specific  for a  particular strain of £,. coli.
 the  problematic tropical  background  flora  are eliminated.   The only
 problem is  that if E.. colj is  able  to  survive for prolonged periods  in
 tropical  waters, the viruses that infect them could do  so  too.   Coliphage
 assays  would  only  be appropiate  in  tropical  environments if  it was
 demonstrated  that only one strain of £.. coli  is not capable of
 extraenteral survival (Hazen,  1988).  Studies  to date  (Elias  et  al.,  1988;
 Lopez et al.,  1987;  V aide's-Collazo et al., 1987; Hazen  et  al., 1981;
 Jimenez et  al.,  1988;  Perez-Rosas  et  al.,  1984; Carrillo et  al.,  1985) have
 shown this  to  be  unlikely, since all tested strains of E. coli  were  capable
 of  extraenteral survival  in tropical  waters.
      We  have already  mentioned the  phenomena of  viable  but
 unculturable bacteria.  This  has  been  show  to  happen with many
 pathogens  in  both  temperate and  tropical waters,  which suggests that
 most of the time,  indicators  may not be  correlated to disease  risk
 (Baker et al.,  1983;  Colwell et al.,  1985;   Hazen et  al.,  1987).   Considering
 this,  the best  way to assess  tropical  source water  might  well  be  the
 direct enumeration  of the pathogens themselves.  Several  approaches  to
 pathogen  enumeration  have  been proposed.   Among  these,
immunofluorescent  staining have  been  shown to  detect   extremely low
HO cells/ml) levels  of pathogenic  bacteria (Colwell et al., 1935).   Cross

-------
                                                                           52
reactivity,  however, hampers  the potential of  this approach, even  -A hen
monoclonal antibodies  are  used.
      Currently, DNA, cDNA, or RNA probes may be the most sensitive
method for pathogen detection.   DNA  probes have  been developed for
Salmonella sp., and for enterotoxigenic  £.. coli (Hazen, 1938). The
detection  of  any of these in  tropical drinking  water  would  surely point
to an  imminent  human health  hazard.

-------
                                                                             53
                             CONCLUSIONS

1.   Turbidity  is  a  very important factor affecting  water  treatment.   By
     associating  themselves  with  paniculate  matter, bacteria gain
     protection  against  chlorine  disinfection.   Whenever  water  turbidity
     increases,  special  care  should be taken  in  analyzing  samples to
     determine  bacterial  water quality.
2.   Turbidity  also  affects the  performance  of the media  used for the
     detection  of  indicator bacteria.   The high levels of background
     bacteria brought  about  by  high turbidity  of  both  influents  and
     effluents can  interfere  with  the  expression  of the characteristics
     normally  associated  with  indicator  bacteria  on  differential  media.
     High  turbidity  also  clogs  membrane filters and makes it difficult  to
     observe bacterial  growth  on  fermentation tubes.
3.   Coliforms  may  survive  treatment  and  remain in  an  unculturable,
     but viable  (injured) state.   Injured  coliforms cannot  be  detected
     using  standard  methodology.   This  can  result in  water being
     classified as potable when  it  is  not.   The only method capable of
     detecting  injured  coliforms  was  membrane  filtration   using  m-T7
     agar.
4.   Membrane  filtration  using  m-T7 agar  recovered  less nontarget
     bacteria than any  other  fecal coliform  method.  This  method  could
     be  used to  recover  naturally occurring  coliforms  with higher
     selectivity  in  tropical  environments.
5.   Fecal  streptococci are inadequate indicators  of water  quality  in  the
     tropics.   Some components  of  the  normal  environmental flora are
     able to  give  false positive fecal  streptococci  reactions  on standard

-------
                                                                             54
    media.   In  addition, further  investigation is needed  to  determine
    the  higher  resistance of  fecal streptococci  to  chlorine disinfection
    and the  occurrence  of  tropical fresh water  bacteria  capable of
    growing  on culture  media  containing  sodium  azide7
6.  Although  standard  water treatment  was  able  to significantly reduce
    the  total  bacterial densities  present in source water,  AODC was able
    to  detect high  concentrations  of  bacteria in drinking water.   Further
    research  as  to  the  nature and identity of this background flora  is
    needed.
1.  Due to the  high  numbers of  fecal coliform positive bacteria of non-
    fecal  origin, water  treatment  plants  in  Puerto  Rico are
    overchlorinating finished  drinking  water.   Thus, not only  is  the
    chlorination  process  treating a health risk, it is  creating one  by
    increasing the  concentration  of  trihalomethanes in the  water.
8.  In  terms  of bacterial densities, raw  water brought in  from Cidra
    Lake  to  the Guaynabo  treatment  plant  had unique  characteristics
    relative to those  of  the  other  plants  studied.   The possible effects of
    water  travel through  such a  long  pipeline,  in  terms  of bacterial
    quantity  and quality,  should  be  further  studied.
  9.   New,  better  indicators of tropical  drinking  water  quality are  an
      immediate necessity.   The  only  practical  solution  to  this problem
      might  be  the direct isolation of pathogenic  microorganisms from
                               drinking water.

-------
                                                                            55
                            LITERATURE  CITED

American Public Health Association.   1984,  Standard Methods for the
      Examination  of Water  and Wastewater,  16«i>  edition. American
      Public Health Association,  Inc.  Washington, D.C.
Badge, U. S., A. M. Khan, and A. K. Varma.  1982.  Influence of physico-
      chemical  factors  on the coliform bacteria in a  closed-lake water
      system.   Inter. J. Environ. Stud.  18:237-241.
Baker, R. M., F. L. Singleton, and M. A. Hood.  1983.  Effects of nutrient
      deprivation on  Vibrio cholerae.   Appl. Environ.  Microbiol. 46:930-
      940.
Barrel, R. A. E,, and M. G. M. Rowland.  1979.  The relationship between
      rainfall and well  water pollution in a West  African  (Gambian )
      village.  J. Hyg.  Camb. 83:143-150.
Berg, G., and T. G. Metcalf.  1978.   Indicators of Viruses in Waters.  In
      "Indicators of Viruses in Water and Food".  G.  Berg  (ed.),  Ann
      Arbor  Sciences.
Sergey's  Manual of Determinative Bacteriology.   1974.  Buchanan,  E.,
      and N. E. Gibbons, eds.  The  William Wilkins Company, Baltimore,
      MD.
Bermudez, M., and T. C. Hazen.   1988.  Phenotypic and genotypic
      comparison of  Esche.richia coli from  pristine tropical waters.   AppL
      Environ.  Microbiol. 54:979-983.
Biamon,  E. J., and  T. C. Hazen.  1983.  Survival and distribution  of
      Aeromonas   hydrophila  in  near-shore coastal waters  of  Puerto  Rico
      receiving  rum  distillery  effluent.   Wat.  Res.17:319-326.

-------
                                                                            56
 Bigger, J. W.  1937.  The growth of coliform  bacilli in water.  J.  Pathol.
      Bacteriol.  44:167-211.
 Bisson, J. W., and B. J. CabellL  1980-  Clostridium perfringens as a water
      pollution indicator.  J. Water Pollut. Control Fed. 52:
 Bonde, G. J.   1977.  Bacterial indication  of water pollution.   Adv. Aquatic
      Microbiol.  1:273-364.
 Burlingame, G. A., J, MacElhaney, M. Bennet, and W.  O. Pippes.  1986.
      Bacterial  interference  with coliform colony sheen  production  in
      membrane  filters.  Appl.  Environ.  Microbiol.  47:56-60.
 Cabelli,  V.  J.,  1983.  Public  health  and  water quality significance of viral
      diseases  transmitted  by  drinking  water  and recreational  water.
      Wat.  Sci. Tech. 15:1-15.
 Camper, A.  K., M. W. LeOievallier, S. C.  Broadaway,  and G.  A.  McFeters.
      1986.   Bacteria  associated  with  granular activated carbon particles
      in drinking water.  Appl. Environ.  Microbiol.  52:434-438.
 Carrillo, M., E. Estrada, and T. C. Hazen.  1985.  Survival and enumeration
      of the fecal indicator Bifidobacterium adolescentis and Escherichia
      coli in  a tropical rain forest  watershed.   Appl.  Environ.  Microbiol.
      50:468-476.
 Clark,  J.  A.  1980.   Influence of increasing  numbers of non-indicator
      organisms  upon  detection of  indicator organisms in  the  membrane
      filter  and presence absence tests.   Can. J.  Microbiol. 26:465-470.
Clark,  T. F.  1984.   Chlorine tolerant bacteria  in a water distribution
      system.   Public  Works  115:65-67.
Colwell, R. R., P.  R. Brayton, D. J. Grimes, D. B. Roszak,  S. A.  Huq, and L. M.
      Palmer.   1985.   Viable  but non-culturable  Vibrio cholerae and

-------
      related  pathogens  in  the  environment:implications  for  release  of
      genetically  engineered organisms.   Biotechnology.  3:817-820.
Curtis, R. E., S. Guzman-Rfos, and  P. L. Diaz.   1984.  Water Resources
      Data:Puerto Rico and  the Virgin Islands.   U.S. Geological  Survey
      Water  Data  Report, PR-84-1.
Difco Laboratories.  1984.  Difco Manual:Dehydrated Culture Media and
      Reagents for Microbiology; 10th ed.  Difco Laboratories; Detroit, MI
Dutka, B. J., and K. K.  Kwan.  1978.  Health  indicator bacteria in water
      surface  microlayers.   Can. J. Microbiol.  24:187-188.
Elias, E. E., A. Calvo, and T. C. Hazen.  In Press.   Survival and ditribudon
      of Yersinia enterocolitica in tropical freshwater  and food.  Curr.
      Microbiol.
Evans, T. M.,  C.  E. Warwick, R. J.  Seidler, and  M. W. LeChevallier.  198la.
      Failure  of  the  most-probable-number  technique  to detect
      coliforms  in drinking  water  and  raw water  supplies.   Appl.
      Environ. Microbiol.  41:130-138.
Evans, T. M.,  M. W. LeChevallier,  C E. Waarvick, and R. J. Seidler.   198Ib.
      Coliform  species  recovered  from untreated  surface water and
      drinking  water  by  the membrane  filter, standard,  and modified
      most probable  number techniques.   Appl. Environ.  Microbiol.
      41:657-663.
Evison,  L. M., and A.  James. 1975.  Bifidobacterium as an indicator
      ofhuman  fecal pollution in water.  Prog. Water  Technol.  7:57-66.
Feachem,  R.   1976.  An improved role for fecal coliform  to fecal
      streptococci  ratio  in  the  differentiation  between  human  and non-
      haman  pollution  sources.  Water. Res.   9:689-690.

-------
                                                                            53
Feachem, R.   1974.   Faecal  coliforms and  faecal streptococci in streams
      in the New Guinea  highlands.  Wat.  Res.  8:367-374.
FranzblauT S. G., B.  J. Hinnenbausch, C. M. Kelly, and N. A. Sinclair.  1984.
      Effect of noncoliforms on  coliform detection in  potable
      groundwater:  Improved  recovery  with  an  anaerobic  membrane
      filter technique,   Appl. Environ. Microbiol. 48:142-148.
Fuentes, F.  A., E. J. Biamo'n,  and  T.  C. Hazen.  1983.  Bacterial chemotaxis
      to effluent  from a  rum  distillery  in  tropical near-shore coastal
      waters.   Appl.  Environ.  Microbiol.  46:1438-1441.
Fujioka, R.  S.,  H.  H. Hashimoto, E. B. Siwak, and H. F. Reginald.  1981.
      Effects of sunlight  on  survival of  indicator bacteria in seawater.
      Appl.  Environ.  Microbiol.  41:690-696.
Fujioka, R.  S., and L.  K.  Shizumura.  1985.  Clostridium perfringens. a
      reliable  indicator of stream water  quality.   J.  Water Pollut. Control
      Fed.  57:986-992.
Geldreich,  E. E., and B. A. Kenner.  1969.   Concepts  of  fecal  streptococci in
      stream pollution.  J. Water Pollut.  Control  Fed.  41:R366-R352.
Goyal,  S. M.,  and C. P. Gerba.   1983. Viradel method  for detection of
      rotaviruses  from seawater.   J. Virol.  Meth. 7:279-235.
Goytia,  J. R.   1978.  El cloro y su importarscia en  la desinfecci<5n del agua
      que  consumimos.   Boletfn  de Asesoria T&nica,  Autoridad de
      Acueductos  y Alcantarillados* Estado  Libre  Asociado  de  Puerto
      Rico.
Grabow, W. 0. K., C.  A.  Hilner, and P. Coubrough.  1981.  Evaluation of
      standard  and  modified  m-FC,  McConkey, Teepol  media  for
      membrane  filtration counting  of fecal coliforms  in  water.   Appl.
      Environ.  Microbiol.  44:192-199.

-------
                                                                            59
Hagler,  A.  N., L. C. Mendoca-Hagler, E. A. Santos, S.  Farage, J. B.  Silva
      Filho, A.  Schrank, and R. B. De Oliverira.  1986.   MicrobiaJ pollution
      indicators  in Brazilian  tropical and  subtropical marine  surface
      waters.   The Science of the  Total  Environment 5& 151-160.
Hajna, A.  A., and C.  A.  Perry.  1943.  Comparative study of presumptive
      and  confirmative media  for  bacteria of the coliform group and  for
      fecal streptococci.  Am.  J.  Public  Health  33:550-556.
Hazen,  T.  C,, and C. F. Aranda.  1981.   The relationship between the
      distribution and  abundance of  bacteria and  water quality  in  the
      Rio  Mameyes watershed,  pp.  87-111, 7t& Annual Natural
      Resources  Symposium  of Puerto Rico,  Dept.  of Natural Resources.
      Commonwealth  of Puerto Rico, San Juan.
Hazen, T.  C.,  L.  Prieto, A. Lopez, and E.  Biamon.  1982.  Survival and
      activity of fecal coliform bacteria  in  near-shore coastal  waters.8th
      Annual  Natural  Resources  Symposium of Puerto Rico, Dept. of
      Natural  Resources.   Commonwealth of Puerto Rico.
Hazen, T.  C.,  J. Santiago-Mercado,  G.  Toranzos,  and  M.  Bermudez.  1987.
      What does the   Presence  of Fecal Coliforms Indicate  in the Waters
      of Puerto  Rico?:A Review.  Bol. Asoc. Med. P. Rico, Mayo 1987:193-
      193.
Hejkal, T.  W., F. M. Weilings, P. A.  LaRock,  and A. L. Lewis.  1979.
      Survival  of poliovirus within  organic  solids  during chlorination.
      Appl. Environ.   Microbiol. 38:114-118.
Hejkal, T. W.,  R. L. LaBelle, B.  Keswick, C. P. Gerba,  B. Hofkin, and Y.
      Sanchez.   1982.   Viruses  in  a  small  community  water supply
      associated  with  a gastroenteritis epidemic.   J. Am.  Water  Works
      Assoc.  74:318-321.

-------
                                                                            60
Hill, M.  J.,  B. S. Drasar, V. Aries, J. S. Crowether, G. Hawksv,orth, and
      R.E.O. Williams.  1971.   Bacteria and  Aetiology  of Cancer of Large
      Bowel.   Lancet  1:95-100.
Hill, G., and H.  Rai.   1982.  A preliminary  characterization of the tropical
      lakes of the central  Amazon by  comparison  with polar and
      temperate  systems.   Arch.  Hydrobiol.  96:97-111.
Hoadley, A. W., G. Ajillo, and N. Mastherson.  1975.   Preliminary studies
      of fluorescent Pseudomonas  capable of  growth at 41 °C  in
      swimming  pool  water.   Appl. Microbiol.  20:527-531.
Hobbie, J.  E., R. J. Daley, and  S. Jasper.  1977.  Use of Nuclepore  filters for
      counting bacteria  by fluorescence  microscopy.   Appl.  Environ.
      Microbiol.  33:122-1228.
Hoff, J.  C   1978.  The  relationship  of turbidity to  disinfection of potable
      water, p.  103-117.   In C. W. Hendricks  (ed.). Evaluation of the
      microbiology  standards  for  drinking  water.   Report EPA-570/9-
      78-OOC.  U.  S.  Environmental Protection Agency. Washington,  D. C.
Hutchinson, M., J,  and W.  Ridgway.   1977.  Microbiological aspects of
      drinking  water supplies.   In  Aquatic Microbiology.   F.  A.  Skinner
      and J. M.  Sheman, editors.   Academic Press,  London.
Jacobs, N. J., W. L. Zeigler, F. C. Reed, T. A.  Stukel,  and E. W. Rice. 1986.
      Comparison  of   membrane   filter, multiple-fermentation-tube,  and
      presence-abscence  techniques for  detecting  total  coliforms in
      small community  water  systems.   Appl. Environ. Microbiol.
      51:1007-1012.
Jimenez,  L., I. Muniz,  G. A. Toranzos,  and T. C. Hazen.  In  Press.   The
      survival and  activity  of  Salmonella tvphimurium and Escherichia
      colj  in tropicaJ  freshwater. J. Appl. Bacteriol.

-------
                                                                             &
Kenner, B, A., H. F. Clark, and P. W. Kabler. 1961  Fecal
      streptococci:Cultivation and enumeration  in  surface  waters.   Appl.
      Microbiol.  9:15.
Keswick, B. H., C. P. Gerba, J. B. Rose, and G. A. Toranzos.  1985.  Detection
      of rotaviruses in treated  drinking  water.   Water  Sci. Technology.
      17:1-6.
Lavoie, iM.  C.   1983,   Identification of strains isolate-d as total and fecal
      coliforms  and comparison  of both groups  as indicators  of fecal
      pollution in tropical climates.  Can.  J.  Microbiol. 29:689-693.
LeChevallier, M. W., S. C. Cameron, ajid  G.  A. McFeters.  1983.  New
      medium for  improved  recovery  of coliform bacteria from  drinking
      water.  Appl. Environ. Microbiol. 45:484-492.
LeChevallier, M.  W., and G.  A,  McFeters.   1985.  Enumerating  Injured
      Coliforms  in  Drinking  Water.  J. Amer. Wat. Works Ass. June
LeChevallier, M. W., T. M. Evans,  and R. J.  Siedler.  1981.  Effect of
      turbidity  on  chlorination   efficiency  and  bacterial persistence  in
      drinking  water.   Appl. Environ.  Microbiol.  42:159-167.
UChevallier, M. W., C. D. Cawthon, and  R. G. Lee. 1988.  Factors
      promoting  survival of  bacteria  in  chlorinated   water  supplies.
      Appl. Environ.  Microbiol.  54:649-654.
Levy, R. V., R. D. Cheetham, J. Davis, G. Winer,  and F. L. Hart.  1984.
      Novel method for studying  the  public  health significance  of
      macroinvertebrates  occurring in  potable water.    Appl.  Environ.
      Microbiol.  47:889-894.
Lopez-Torres, A. J.   1982.  Ecology of Klebsiella pneumonias,  in  aquatic
      habitats of Puerto Rico.  M.S. Thesis.   University of Puerto Rico, Rio
      Piedras.   90 p.

-------
                                                                           62
L6p€z-Torres, A. J., T.  C.  Hazen, and G. A. Toranzos.  1987.   Distribution
      and  in situ  survival  and activity  of  Kiebsielia pneumonias and
      Escherichia coll in  a tropical rain forest watershed.   Curr.
      Microbiol.  15:213-218,
MacFaddin, J. F.   1980. Fecal boths (KF-SF) for  enterococci.  In
      Biochemical Tests for the  Idntification  of Medical Bacteria.  Chap.
      11,  pp.  114-120.
McFeters, G. A., J. S. Kippin, and M. W. LeChevallier.   1986.  Injured
      coliforms  in drinking  water.   Appi. Environ.  Microbiol. 51:1-5.
McFeters, G. A., and D. G. Stewart.  1972.  Survival of coliform bacteria in
      natural  watersrfield  and laboratory  studies with MF  chambers.
      Appl. Environ. Microbiol.   24:805-811.
Means,  E.  G., and B. H. Olson.   1981.  Coliform  Inhibition  by Bacteriocin-
      Like  Substances  in  Drinking  Water  Distribution  Systems.  Appl.
      Environ.  Microbiol.    42:506-512.
Muniz,  L, G. A. Toranzos,  and T. C. Hazen.  In Press.  Survival and
      activity of Streptococcus faecalis  in a  tropical  rain  forest stream.
      Appl. Environ. Microbiol.
National Academy of Sciences.   1977.  Summary Report:  Drinking  Water
      and Health.  A Report of  the Safe Drinking Water Committee,
      Advisory  Center  on  Toxicology,  Assembly  of  Life Sciences, National
      Research  Council. Washington, D.C., 1977.
Odum, E. P.  1971.  Fundamentals  of Ecology.  W. B. Saunders Company.
      Philadelphia.   574  pp.
Oluwande, P. A., M. K. C.  Sridhar, A. O. Bammeke, and A. O.  Okubadejo.
      1983.  Pollution  Levels in  some  Nigerian Rivers.  Water  Res.
      17:957-963.

-------
                                                                            63
Ortiz-Roque, C,, and T. C. Hazen,  1983.   Legionella spp. in  the waters  of
      Puerto Rico and a  short review  of  legionellosis.  Bull.  Puerto  Rico
      Med. Assoc.  75:403-407.
Pagel, E. J.,  A. A. Queschi, M. D. Young, and L. T. Vasoff^ 1982.
      Comparison of  four  membrane  filtration methods   for fecal
      coliform enumeration.   Appl. Environ. Microbiol.  43:787-793.
Peele, E. R., F. L. Singleton, J. W. Deming, B. Cavari, and R. R. Coiwell.
      1981.   Effects  of  pharmaceutical  wastes on microbial  populations
      in surface  waters  at  the Puerto  Rico  dump  site  in  the  Atlantic
      Ocean.   Appl. Environ.  Microbiol. 41:873-879.
Perez-Rosas, N., and T.  C.  Hazen.  1988.   In  situ survival  of Vibrio
      cholerae and Escherichia coli in  tropical coral reefs.  Appl. Environ.
      Microbiol.  54:1-9.
Ptak, D. J., W. Ginsburg, and B.  F. Willey.  1973.   Identification and
      incidence of  Klebsiella  in chlorinated  water supplies.  J.  Am. Water
      Works  Assoc. 65:604-608.
Puerto  Rico Department of Health.  1987.  Comunicado  de Prensa.
Puerto  Rico Department of Health.  1987.   Resumen  Ejecutivo del  Esrudio
      Epidemiologico  del  Brote  de Gastroenteritis en  Barrios y Sectores
      de Ranchera, Collores y Algarrobo-Cafiita de Yauco, Puerto Rico.
      Department  of Health,  Commonwealth  of Puerto Rico.
Puerto Rico  Department of Health.  1986.  Sanitary Survey  Report of
      Public Water  Supply  Systems: La  Plata System.   July  23,  1986,
      Department of Health,  Commonwealth  of Puerto Rico.
Puerto Rico  Department of Health.  1985.  Sanitary Survey  Report of
      Public Water  Supply Systems:  Villalba Urban System.   August  12,
      Department of Health,  Commonwealth  of Puerto Rico.

-------
                                                                           64
Puerto Rico  Department of Health.   1987.   Sanitary Survey  Report  of
      Public  water Supply  Systems:Yauco's  Ranchera System.   July  13-
      1987.  Department of Health, Commonwealth of Puerto Rico.
Puerto Rico  Department of Health,  Environmental Health- Division.   1987.
      Filtration  Plant  Inspection  Report:Rancheras Water System:July
      1987.  Department of Health, Commonwealth of P-uerto Rico.
Puerto Rico  Environmental  Quality Board.   1985.  Puerto Rico Virus,
      Parasite  and Bacteria Study  Projecf.Initial  Comprehensive  Report
      Compiled  by the  Environmental  Quality  Board.  Office of the
      Governor/Environmental  Quality Board,  Commonwealth of Puerto
      Rico.
Ragavachari,  T. N. S., and P. V. W.  Iyer.  1939.  Longevity of coliform
      organisms in water  stored under  natural conditions.   Indian J.
      Med.  Res.  26:877-883.
Ray. B.,  and  M. Specks.  1973.  Discrepancies in  the enumeration of E.
      coli.  Appl. Microbiol. 21:46-49.
Reilly, J.  K., and J. S. Kippen.  1983.  Relationship of bacterial counts with
      turbidity  and free chlorine in two  distribution systems.  J.  Am.
      Water  Works  Assoc.  75:309-312.
Rivera, S. C, T. C. Hazen, and  G. A. Tonmzos.  1988.  Isolation of fecal
      coliforms from  pristine sites  in  a  tropical rain forest.  Appl.
      Environ.  Microbiol, 54:513-517.
Rodriguez, L. , I. Perez-Suarez, M. M. Rivera,  and  S. File.  1988.   Detection
      of Cryptosporidium sp and Giardia sp. in  surface  water and
      sewage in tropical  areas.   Abstracts of  the  Annual Meeting of the
      American Society for Microbiology.  N-19,  p. 33.

-------
                                                                            65
Rose, R. E., E.  E.  Geldreich,  and  W. Litsky.   1975.  Improved  membrane
      filter  method for  fecal  coliform  analysis.   Appl. Microbiol.  29:533-
      536.
Roszak, D. B., D. J. Grimes, and R. R. Colwell.  1984.  Viable but
      nonrecoverable  stage of  Salmonella enteriditis in aquatic  systems.
      Can.  J. Microbiol. 30:334-338.
Ruiz  de  la  Mata, B.   1985 (8/26).  Lake Canaizo's Deceptive  Picture.   San
      Juan Star.  pp. 1 &  16.
Santiago-Mercado,  J., and  T. C. Hazen.  1987.  Comparison of  four
      membrane filter methods  for  fecal coliform  enumeration in
      tropical waters.   Appl.  Environ.  Microbiol.  53:2922-2928.
Silverman, G. S,, L. A. Nagy,  and  B. H. Olson.  1983.   Variation in
      paniculate matter, algae,  and bacteria  in  an uncovered finished
      water  reservoir.   J. Am. Water  Works  Assoc.  75:191-195.
Thomson, J. A.   1981.  Inadequacy of  Escherichia coli as an  indicator  of
      water  pollution in a tropical climate:A  preliminary study in
      Botswana.  South African Journal  of  Science 77:44-45.
Toranzos, G. A., and C. P. Gerba.   1988.  Development of an efficient
      method for  concentration  of rotaviruses from  large volumes of
      water. J. Virol. Meth.   In press.
Toranzos, G. A., C. P. Gerba,  M. Zapata, and F. Cardona.  1986. Presence
      de virus  enteriques  dans des eaux  de  consommation  a
      Cochabamba  (Bolivie).   Revue  Internationale  des  Sciences  de 1'eau.
      2:91-92.
Tracy, H. W,, V. M. Camarena, and F.  Wing.  1966.  Coliform persistence
      in  highly  chlorinated waters.   J. Am.  Water Works Assoc.  58:1151-.
      1159.

-------
                                                                           66
United  States  Environmental  Protection Agency,  Caribbean Field Office,
      Water  Management Division.   1987.   Non-PRASA  Public  Water
      Systems Report.
United States Water Resources Council.  1978.  Caribbean- Region.  U.S.
      Water  Resources  Council, Washington, D.C,
Valdes-Collazo, L., A. J. Schultz, and T. C, Hazen.  1987.  Survival of
      Candida a 1 b i c a n s in  tropical  marine  and fresh  waters.  Appl.
      Environ.  Microbiol.  53:1762-1767.
Wheater, D.  W. F., D. D. Mara, and J Oragui.  1979.  Indicator systems to
      distinguish  sewage from  stormwater  run-off  and  human  from
      animal  fecal material.   In Biological  Indicators of Water  Quality.
      eds.  James,  A., and Evison, L.  John Wiley & Sons,  Toronto, Canada,
      pp.  21/1-21/27.
World Health Organization.    1987,  Guias para la calidad  del agua
      potable.  Vol. 2.  World Health  Organization.
Wright, R. C.  1982a.  A comparison  of the levels of faecal indicator
      bacteria  in  water and  human  faeces in a rural  area of  a tropical
      developing Country (Sierra Leone).  J. Hyg. Camb.  89:69-73.
Wright, R. C.   1982b.  A medium for the  rapid enumeration  of
      Esc'nerichia coli  in the  presecence of other faecal coliforms in
      tropical waters.   J.  Hyg.  Camb. 88:265-273.
Wright, R.C..  1986.  The Seasonally of Bacterial Quality of Water in a
      Tropical  Developing Country (Sierra Leone).  J. Hyg. Camb,  75-82.
Xu, H. S., N.  Roberts, F. L, Singleton, R. W. Attwell, D. J. Grimes, and  R. R.
      Colwell.   1982.  Survival  and  viability of nonculturable Escherichia
      coli  and Vibrio cholerae  in the estuarine  and marine  environment.
      Microb. Ecol. 8:313-323.

-------
Zar. J.  H.   19S4.   Biostatistical  Analysis.   Prentice-Hall  Inc., Eng!e\vood



      Cliffs, N. J.

-------
                                                                                                           o 8
Table  I.  Water Quality  Parameter by Site  and Water  Source.   (Measurements were  taken  between  September  M,
         1987  and March 7, 1988).
INFLUENTS
Site
Bayam6n
Guaynabo
Villalba
Yauco
EFFLUBTO
Site
Bayamon
Guaynabo
Villalba
Yauco
Temperature
CO
1
24.45 ± 0.39
24.00 ± 0.63
22.50 ± 0.23
21.46 ± 0.37

Temperature
TO
25.18 ± 0.30
25.00 ± 0.39
22.67 ± 0.22 '
23.15 ± 0.32
PH
7.76 ± 0.13
7.36 ±0.17
7.86 ± 0.18
7.96 ± 0.10

PH
7.79 ± 0.17
7.20 ±0.16
7.57 ±0.19
7.73 ± 0.19
TCI'
(mg/L)
0.02 ± 0.02
0.00 ± 0.00
0.00 ± 0.00
0.00 ± 0.00

TCP
(mg/L)
1.13 ± 0.08
1.34 ± 0.17
2.41) ±0.15
1.43 ± 0.09
FCI2
(mg/L)
0.00 ± 0.00
0.00 ± 0.00
0.00 ± 0.00
0.00 ± 0.00

FC12
(mg/L)
0.96 ± 0.06
0.68 ± 0.10
1 68 ± 019
0.89 ± 0.08
Turbidity
(NTU)
34.03 ± 9.61
19.48 ± 8.77
60.75 ± 30.00
6.57 ± 1.89

Turbidity
(NTU)
1.60 ± 0.28
1.19 ± 0.12
1 99 ± 0 20
1.12 ± 0.14
All  values are mean ± one standard  error,  1  = Total chlorine, 2 = Free Chlorine.

-------
I able 2.  Bacteriological Density by Site  and Water Source. (Measurements  were taken between  September 24.

         and  March 7,  1988).

INPUJENTS
m-Endo
Site
Bay a rn 6 n
Guaynabo
Villalba
Yauco
(CPU)
1074 ±
3662 ±
28475
1808 ±

553
1835
± 2030
859
m-FC
(CPU)
190 ±


84
1510 ± 667
903 ±
233 ±
354
89
m-T7
(CPU)
80 ± 46
439 ± 265
146 ± 45
127 ± 77
m-KF
(CPU)
203 ± 68
1928 ± 1006
950 ± 471
948 ± 333
MPN1



133 ± 79
1285 ±
1068 ±
949 ±
166
172
204
MPN2

62 ± 30
969 ± 1
984 ± 1
564 ± 1



98
77
76
MPN 3

94 i 4 A
1 1 23 1 !
1 1 2(> ± 1
930 + 20
EFFLUENTS
m-Endo
Site
Bayam6n
Guaynabo
Villalba
Yauco
(CPU)
0.00 ±
0.25 ±
0.75 ±
0.62 ±

0.00
0.18
0.79
0.37
m-FC
(CPU)
0.00 ±
0.00 ±
2.00 ±
0.54 ±


0.00
0.00
2.00
0.37
m-T7
(CPU)
1.18 ± 1.18
O.(X) ± 0.00
0.45 ± 0.45
0.09 ± 0.09
m-KF
(CPU)
6.36 ± 5.40
0.17 ± 0. 1 1
1.92 ± 0.79
4.31 ± 3.46
MPN1

2.00 ±
2.92 ±
2.00 ±
2.46 ±


0.00
0.92
0.00
0.46
MPN 2

2.00 ±
3.75 ±
2.50 ±
2.00 ±


0.00
1.75
0.50
0.00
MPN 3

2.91 d
12.7 d
2.00 J
2.46 d
All  values arc mean ± one  standard error per 100 ml.  MPN1 = Total coliforrn  MPN.  MPN2 =  Pecal coliform MPN.
= Fecal streptococci MPN

-------
Table 4.   Average Indicator Organism / AODC* Ratios per Viable Count Method Assessed

         (Expressed  as  Percentages).

INFLUENTS
                                                                                                         7 I

SITE

Bayam6n
Guaynabo
Villalba
Yauco
m-Endo
agar
i
7.29
313
96
98
m-FC
agar

2.63
128
15.7
12.9
m-T7
agar

1.21
33.7
2.39
2.50
m-KF
agar

3.64
108
20.2
66.4
MPN1


2.25
191
19.8
43.0
MPN2


9.24
166
19.6
26.0
MPN3


1.56
189
21.0
68.0
EFFLUENTS

SITE
Bayamon
Guaynabo
Villalba
Yauco
m-Endo
agar
0.00
0.00
0.02
0.05
m-FC
agar
0.00
0.00
0.07
0.03
m-T7
agar
0.07
0.00
0.02
0.00
m-KF
agar
0.33
0.04
0.06
0.38
MPN1

0.14
3.27
0.15
0.28
MPN2

0.14
4.60
0.24
0.27
MPN3

0.19
19.0
0.15
0.28
All  valures  are percent X  10-4,  MPNl=Total coliform  MPN,  MPN2=Fecal coliform MPN,  MPN3=
      Fecal streptococci MPN.

-------
                                                                                                         7 0
Table 3.   Direct  Count and Percent Activity  by Site  and  Sample Type  (Measurements taken between Scptemlx-r  M.

         1987  and March 7,  1988)
                 BAYAMON                                                         GUAYNABO
S a m p 1 c
Type
Influent

Effluent
Sample
Type
Influent

Effluent
AGDC*
(X 107)
7.19 ± 3.67

1.74 ± 0.44
                 VILLALBA
AODC*
(X 107)
3.44  ± 0.46

1.42  ± 0.27
Percent
Activity**
73.84 ± 4.15

56.31 ± 3.58
Percent
Activity**
76.38 ± 3.65
62.69 ± 6.27
Sample
Type
Influent
Effluent
AODC*
(X 107)
4.39 ± 0.76
0.82 ± 0.41
Percent
Activity1"*'
32.14 ± <>.?.
26.53 i 5.6
YAUQO
Sample
Type
Influent
Effluent
AODC*
(X 107)
5.02 ± 0.96
2.53 ± 0.38
Percent
Activity**
50.45 ±
36.5K ±
*Acridir»e  Orange Direct Counts.
**as determined  by the number of  red fluorescing cells divided by the total number of fluorescent cells and
   multiplied  by  100.

-------
Table 5.   Correlation  Matrix: Combined  Influent Bacteriological  Methods  and  Water Quality
MFTC
MI: TC
MF FC
MF MT7
MF FS
MPNTC
MPN FC
MPN FS
AdX
%ACT.
TEMP
PH
TURB
1
0
0
0
0
0
0
0
0
0
0
0
.000
.889***
i
.708**
.818**
.828***
.798**
.766**
.083
.052
.218
.117
.064
MFPC
1
0
0
0
0
0
0
0
0
0
0
.000
.699**
.724**
.738**
.732**
.687**
.004
.101
.185
.175
.083
MFMT7
!
0
0
0
0
0
0
0
0
0
.000
.552
.594*
.619*
.648*
.042
.128
.223
.049
.098
MFFS

1.000
0.698**
0.706**
0.68*
0.097
0.100
0.027
0.241
0.140
MPNTC


1.000
0.884***
0.888***
0.167
0.216
0.233
0.135
0.056
MPN FC



1.
0.
0.
0.
0.
0.
0.



000
902***
127
117
314
147
020
MPN FS ACOC %ACT TEMP PH TUKH




1.000
0.232 1.000
-0.146 0.345 1.000
0.253 0.124 0.028 1.000
-0.108 0.206 0.039 0.440 1 .000
0.01 1 0.380 0.435 0.090 0.033 1.000
n=44, * = P < 0.05, ** - P < 0.01. *** » P < 0.001. MF TC - m-Endo agar, MF FC - m FC agar. MF MT7 - m-T7 «g«r. MF FS - ra-KF a^ar. MPN
Total coliform MPN. MPN FC « Fecal coliform MPN. MPN FS = Fecal streptococci  MPN. AODC » Acridine Orange Direct Counts. %ACT  =
Percent activity,  TEMP » Temperature.  TURB = Turbidity.

-------
Table 6.   Correlation  Matrix:  Combined Effluent  Bacteriological  Methods  and  Water Quality.

MFTC
MP FC
MF MT7
MFFS
MPN TC
MPN FC
MPN FS
AODC
%ACT
TEMP
PH
TURB
TCL
FOL
MFTC
1.000
0.839***
0.022 .
0.306
0.23!
-0.077
0.089
0.225
-0.151
-0.032
0.153
0.229
0.220
0.188
MFFC

1.000
0.035
0.221
0.215
-0.051
0.095
0.117
-0.161
-0.066
0.106
0.221
0.098
0.121
MF MT7


1. 000
0.448
0.080
-0.050
0.382
0.097
0.031
0.038
0.106
0.540
0.041
0.021
MFFS


8.000
0.084
-0.052
0.247
0.215
0.163
-0.107
0.353
0.478
0.152
0.184
MPNTC


1.000
0.573
0.856*
-0.109
-0.334
0.054
0.067
0.087
-0.136
-0.101
MPNFC


1.000
0.649
-0.279
-0.192
0.036
-0.008
0.247
-0.690
-0.036
MPN FS



1.000
-0.162
-0.247
0.115
0.113
0.316
-0.1 13
-0.093
AOUC



1.000
0.341
-0.058
0.069
0.148
0.283
0.268
%ACT TEMP PH TURB TU . I-C




1.000
-0.102 1.000
0.192 -0.262 1.000
0.277 -0.112 0.195 1.000
0.446 -0.484 0.024 0.334 1 .()()()
0.422 -0.567 0.025 0.321 0.9W*-
n=46, * - P < 0.05, ** - P < 0.01. *** = P < 0.001. T CL = Tola! Chlorine. F CL - Free Chlorine..MF TC - m-Endo agar. MF FC - m-FC aj;»r.
MT7 = m-T7 agar. MF FS - m-KF agar. MPN TC  =  Total coliform  MPN. MPN FC - Fecal coliform  MPN. MPN FS  « Fecal streptococci MPN
= Acridinc  Orange Direct Counts. %ACT =  Percent activity, TEMP  = Temperature. TURB ^ Turbidity.

-------
Table  7.   Correlation  Matrix:  Bayam6n Influent  Bacteriological Methods and Water  Quality.
                                                                                                                            7 4

MFTC
MF F:C
MF MT7
MF f S
MPN TC
MPN I-C
MPN PS
AOCX
%ACT
TEMP
PH
TURB
MFTC
1.000
0.933***
0.742** '
0.625*
0.732**
0.629*
0.616*
0.665*
0.509
0.875***
0.612*
0.028
MFFC
1.000
0.804**
0.53
0.805**
0.652*
0.7*
0.477
0,31
0.881***
0.714**
0.027
MFMT7
1 .000
0.517
0.739**
0.503
0.675*
0.499
0.222
0.718**
0.471
0.089
MFFS

1.000
0.372
0.569
0.316
0.472
0.554
-0.566
0.231
0.247
MPNTC


1 .000
0.771**
0.953***
0.335
0.152
0.624*
0.547
0.131
MPNFC



1.000
0.823***
0.525
0.478
0.66*
0.555
0.264
MPN FS




1.000
0.373
0.134
0.601*
0.628*
0.319
AOUC %ACT TEMP PH TURH





1.000
0.623* 1.000
0.559* -0.548 1.000
0.354 0.24 0.839*** 1.000
0.073 -0.087 -0.036 0.356 1.000
n=l 1, * = P < 0.03, ** - P < 0,01, *** = P < 0.001. MF TC - m-Endo agar, MF FC - m-FC agar, MF MT7 - m-T7 ngnr, MF FS - m-KF agar. MPN
Total coli/orm MPN. MPN FC » Fecal  conform  MPN. MPN FS = Fecal streptococci MPN. AODC « Acridirws Orange Direct Counts, %ACT  =
Percent activity,  TEMP - Temperature.  TURB =  Turbidity.

-------
Table 8.   Correlation  Matrix: Bayam^n Effluent  Bacteriological Methods and Water  Quality.

MFTC
MFFC

MF MT7
MI- FS
MPNTC
MPN FC
MPN FS
AGDC
%ACT
TEMP
PH
TURB
TCL
FCL
MFTC MFFC MF MT7
1.000
1 .000
i
1 .000
0.845***
« a «
» • «
» * *
0.109
0.108
0,056
0.299
0.952***
0.600
0.529
MF FS MPN TC




1. 000
1 .000
» •
0.845*** •
0.095
0.108
0.221
0.234
0.782** •
0.630*
0.662*
MPN FC MPN FS






1.000
1 .000
-0.109
0.108
-0.056
0.299
0.952***
0.600**
0.529
AODC %ACT TEMP PH TURB 1X1, K








1.000
0.431 1.000
0.187 0.252 1.000
0.520 0.080 0.184 1.000
0.054 0.299 0.138 0.262 1.000
0.231 0.316 0.090 0.008 0.733** 1.000
0.415 0.303 0.100 0.137 0.640** 0.927**
n=l I, * = P < 0.05. ** « P < 0.01, *** * P < 0.001, • - Zero values correlation. T CL « Total Chlorine. F CL » Free Chlorine..MF TC = m-Kndo
agar. MF FC » m-FC agar. MF MT7 - nvT7 agar, MF  FS - m-KF agar, MPN TC - Total coliform MPN. MPN FC - Fecal conform MPN. MPN F:
Fecal  streptococci  MPN, AODC  * Acridine  Orange Direct  Counts. %ACT =  Percent activity. TEMP * Temperature, TURB = Turbidity.
ACT = Percent activity;  TEMP = Temperature;  PH = pH;  TURB  =  Turbidity.

-------
Table  9.   Correlation  Matrix:  Guaynabo Influent  Bacteriological Methods and Water  Quality

MFTC
Mr i:c
MF MT7
MF FS
MPN TC
MPN PC
MPN FS
AODC
%ACT
TEMP
PH
TURB
MFTC
1.000
0.934***
0.387 i
0 931***
0.634*
0.516
0.457
0.244
0.473
0.458
0.204
0.79*
MFFC

J.OOO
0.597*
0.796**
0.576*
0.659*
0.508
0.130
0.471
0.605*
0.081
0.656*
MFMT7


1.000
0.376
0.204
0.817**
0.634*
0.553
0.337
0.671*
0.231
0.015
MFFS



1.000
0.286
0.291
0.275
0.459
0.259
0.166
0.080
0.78*
MPNTC




1.000
0.341
0.602*
0.172
0.291
0.686*
0.213
0.254
MPNFC





1.000
0.740**
0.460
0.641*
0.737**
0.239
0.260
MPNFS ACDC %ACT TEMP PH TURK






1.000
0.562 1 .000
0.649* 0.184 1.000
0.687* 0.462 0.169 1.000
0.050 0.318 0.621** 0.290 1.000
0.220 0.618* 0.474 0.030 0.591* 1000
n=l I. * = P < 0.05, ** « P < 0.01, *** = P < 0.001, MF TC = m-Endo agar. MF FC » m FC ngnr. MF MT7 - m-T7 *g«r. MF FS - m-KF a^ar. MF
Total  coliform MPN. MPN FC =* Fccnl  colifomi  MPN,  MPN FS » Fcca!  streptococci  MPN, AODC » Acridinc Orange  Direct  Counts, %ACT
Percent  activity,  TEMP  = Tcmperauire,  TURB =  Turbidity.

-------
Table 10.   Correlation  Matrix:  Guayraabo  Effluent Bacteriological Methods and Water  Quality.
                                                                                                                             77

MFTC
MF FC
MFMT7
MF FS
MPNTC
MPN FC
MPN FS
AOOC
%ACT
TEMP
PH
TURB
TCL
FCL
MFTC MFPC
LOGO
, ! .000
« •
0.194
0.131
0.131
0.131
0.617*
0.258
0.336
0.000
0.202
0.397
0.383
MF MT7 -MF FS


1. 000
1 .000
-0.135
-0.135
-0.135
0.179
0.253
0.173
0.068
-0.316
0.087
0.157
MPNTC




1.000
•
-
0.273
0.249
0.000
0.233
0.371
0.076
0.010
MPNFC





1 .000
•
0.273
0.249
0.000
0.233
0.371
0.076
0.010
MPN FS






1 .000
0.000
0.249
0.000
0.233
0.371
0.076
0.010
ACDC %ACT TEMP PH TURB TU. K







1.000
0.182 1.000
0.438 0.340 1.000
0.342 0.480 -0.370 1.000
0.123 0.407 -0.062 0.528 1.000
0.399 0.481 0.289 0.132 0.481 1.000
0.403 0.406 0.273 0.118 0.523 0.979**
ri=ll. * = P < 0.05, ** - P < 0.01. *** «= P < 0.001, • = Zero values correlation, T CL = Total Chlorine. F CL *» Free Chlorine..MF TC = m-F.ruln
agar, MF FC « m-FC iigar. MF MT7 = rn-T7 agar. MF FS = m-KF ngar. MPN TC = Total colifonm MPN, MPN FC » Fecal colifonm MPN. MPN 1
Fecal streptococci MPN. AODC * Acridine, Orange Direct Counts. %ACT =  Percent  ncttvity. TEMP = Terhperaiurc,  TURB = Turbidity
ACT = Percent activity;  TEMP = Temperaiure;  PH = pH; TURB = Turbidity.

-------
Table 11.   Correlation Matrix:   ViSSalba  Influent Bacteriological  Methods and Water  Quality.
                                                                                                                             7 S

MFTC
MF FC
MF MT7
MF F~S
MPN TC
MPN FC
MJ'N FS
AGDC
%ACT
TEMP
PH
TURB
MPTC
1.000
0.654*
0.739**
0.643*
0.496
0.761**
0.61*
0.138
0.051
0.587*
0.685*
O.H6
MFPC

1.000
0.497
0.331
0.268
0.689*
0.639*
0.080
0.059
0.340
0.554
0.120
MFMT7

i .000
0.4R6
0.296
0.596*
0.76**
0.193
0.364
0.817**
0.525
0.384
MFFS


1 .000
0.547
0.482
0.482
0.330
0.008
0.278
0.271
0.374
MPNTC



1 .000
0.443
0.127
-0.287
-0.268
0.178
-0.454
-0.057
MPNFC




1.000
0.786**
0.419
0.271
0.504
0.410
0.176
MPN FS ACIDC %ACF TEMP PH THRU





1.000
0.340 1.000
0.088 0.577* 1.000
0.653* 0.449 0.069 1.000
0.452 0.119 0.069 0.532 1.000
0.432 0.239 0.490 0.144 0.073 1.000
n=i j. * = P < 0.05, ** - P < 0.01. *** » P < 0.001. MF TC * m-Emlo agar. MF FC » m-FC BRftr. MF MT7 » m-T7 ngar. MJF FS - m-KF agar. MF
Total  colifomi  MPN, MPN FC «•  Fecal  coliforns MPN. MPN FS  =• Fcc«> sircptococci MPN. AODC » Acridine Orange Direct Counts. %AC~I
F'crccni  activity.  TEMP  » Temperature.  TURB  = Turbidity.

-------
Table 12.   Correlation Matrix: Vilialba  Effluent Bacteriological Methods and Water Quality.
                                                                                                                              7 9

MFTC
MF FC
MF MT7
MF FS
MPN TC
MPN FC
MPN FS
AODC
%ACT
TEMP
PH
TURB
TOL
FOL
MFTC
1.000
0.873***
0.128
0.026
«
0.128
•
0.131
0.398
0.363
0.248
0.277
0.335
0.333
MFFC
1 .000
0.091
0.237
a
0.091
•
0.075
0.430
0.539
0.014
0.228
0.030
0.042
MFMT7

1 .000
0.237
•
0.091
•
0.054
0.281
0.135
0.150
0.041
0.269
0.333
MFFS


s.ooo
e
0.004
•
0.255
0.358
-0.419
0.414
0.081
0.447
0.470
MPNTC MPNFC MPN FS



1.000
1 .000
1 .000
0.430
0.258
0.135
0.033
0.588*
0.508
0.457
AODC %ACT TEMP PH TURB K3, K




r

1 .000
0.145 1.000
0.458 0.100 1.000
0.264 -0.237 -0.609* 1.000
0.134 -0.246 0.327 0.07! 1.000
0.244 0.318 -0.355 0.255 -0.403 1.000
0.234 0,290 -0.316 0.194 -0.371 0.981***
n=ll. * = P < 0.05, ** * P < 0.01, *** * P < 0.001, • » Zero values correlation. T CL =* Tolal Chlorine, P CL » Free Chlorine,.MF TC = rn-Hmlo
agar, MF FC = m-FC «gar. MF MT7 == m-T7 agar,  MF FS * m-KF agar. MPN TC - Total coliform MPN. MPN FC « Fccn! coliform MPN,  MPN FS
Fecal  slrcplococci MPN, AODC = Acridine Orange Direct  Counts, %ACT  = Percent  acliviiy.  TEMP  » Tcmpernture,  TURB  = Turbidiiy.
ACT = Percent activity;  TEMP = TefTipcif'ature;  PH = pH;  TURB = Turbidiiy.

-------
Table  13.   Correlation Matrix:   Yaucp  Influent  Bacteriological Method"?  and  Water Quality.
                                                                                                                          X O
MFTC
MFTC
MF FC
MF MT7
MF FS
MPN TC
MPN FC
MPN f^S
AODC
%ACF
TEMP
PM
TUKB
1
0
0
0
0
0
0
0
0
0
0
0
.000
.923***
.658* '
.972***
.915***
.885***
.782**
.411
.029
.696*
.499
.045
MFPC
1.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
000
616*
893***
818***
8 I 3 * * *
694*
440
079
586*
628*
204
MFMT7
1.000
0.615*
0.678*
0.761**
0.724**
0.197
0.401
0.239
0.278
0.242
MFFS

1.000
0.833***
0.807**
0.723*
-0.492
-0.021
0.645
-0.430
0.023
MPNTC


1.
0.
0.
0.
0.
0.
0.
0.


000
98***
851***
241
055
722**
669*
004
MPNFC



1.000
0.835"*
0.136
0.191
0.678*
0.682*
0.099
MPN FS




1.000
0.195
0.021
0.728**
0.581*
0.135
AODC %ACT TEMP PH TURN





1.000
0.732** 1.000
0.404 0.026 1.000
0.18 5 0.123 0.562 1.000
0.558 0.571 0.261 0.227 1.000
n=M, * = P < 0.05. ** - P < 0.01, *** » P < 0.001. MF TC = m-Endo agar, MF FC = m FC agar. MF MT7 » m-T7 agnr. MF FS - m-KF agar. MPN
Total coliform MPN, MPN FC * Fecal col i form MPN.  MPN FS = Fecal streptococci MPN. AODC » Acridinc Orange Direct Counts,  %ACT =
Percent  activity, TEMP » Temperature,  TURB  = Turbidity.

-------
Table 14.   Correlation  Matrix:  Yauco Effluent Bacteriological  Methods and Water  Quality.

MFTC
MF FC
MF MT7
MFFS
MPN TC
MPNFC
MPN FS
AODC
%ACT
TEMP
PH
TURB
TO.
FCL
MFTC
1.000
» i
0.731***
0.903***
0.619*
«
0.681*
0.345
0.255
0.4J6
0.236
0.066
0.342
0.331
MFPC

1.000
0.731**
0.903***
0.619*
•
0.611*
0.345
0.255
0.416
0.236
0.066
0.342
0.331
MFMT7


! .000
0.426
0.893***
•
0.884***
0.420
0.317
0.280
0.015
0.246
0.073
0.010
MFFS



1.000
0.323
«
0.317*
0.158
0.154
0.331
0.374
0.066
0.402
0.403
MPN TC




1.000
•
*
0.584*
0.481
0.263
0.043
0.114
0.204
0.119
MPN FC MPN FS





1.000
1.000
0.589*
0.486
0.261
0.044
0.108
0.209
0.123
AOUC %ACT TEMP PH TURB '1X1, K







1 .000
0.687** 1 .000
0.169 0.010 1.000
0.289 0.206 0.207 1.000
-0.039 0.290 0.273 0.57* 1 .000
-0.326 0.149 0.436 0.187 0.022 1.000
-0.2.40 0.029 0.440 0.216 0.039 0.984***
n=| I. * = P < 0.05. ** » P < 0.0f, *** => P < 0.001, • = Zero values correlation, T CL = Total Chlorine. F CL « Free Chlorinc,.MF TC - m-Hndo
agar. MF FC = ro-FC agar. MF MT7 = m-T7 flgar, MF FS = m-KF agar. MPN TC = Toial coliform MPN, MPN FC = Fecal coliform MPN.  MPN FS
Fecal sircptococcs MPN. AODC = Acridine  Orange Direct  Counts. %ACT =  Percent acliviiy. TEMP •> Temperature. TURB  » Turbidity
ACT = Percent activity;  TEMP * Temperblurtc;  PH =« pH;  TURB *  Turbidily.

-------
Table 15.   Media Performance in the  Assessment  of  Target Bacteria.
INI-UJKNTS
Ayciflfte Recovery*
Endo FC T7
Bayam6n 422
Guaynabo 1912
Villalba 4335
Yauco 1276
144
1087
539
217
66
1 39
99
101
KF
143
558
401
1618
MPN1
133
1256
1019
963
MPN2
61
91 1
928
479
MPN3
94
1079
1086
975
Hi ft K ^* 	 - 	 ---- — 	 	 "> I o w

Endo>FC=T7=KF=MPN 1 =MPN2=M PN 3
Endo=FC=T7=KF=MPN 1 = MPN2=M PN 3
Endo>MPN 1 =MPN2=MPN3>FOT7 =• K I •
Endo=FC=T7=KF=MPNl=MPN2=MJ'N3
0-TLUENTS
AiLCJLag;
Endo
Bayam6n 0
Guaynabo 0
Villalba 0
Yauco 0
c Rccov*
FC
0
0
2
0
?ry*
T7
1
0
0
0
KF
6
0
2
5
MPN1
0
1
0
2
MPN2
1
2
2
0
MPN3
0
1 2
0
1
rcrfoxma.fl.cct

Endo=FC=T7=KF=MPNl=MPN2=MPN3
Endo=FC=T7=KF^=MPN 1 =MPN2=MPN3
Endo=FC=T7=KF^MPN 1 =MPN2=M PN 3
Endo=FC=T7-KF=MPNl=MPN2=MPN3
*Average Recovery =  Average  of  the target colonies recovered on  each  plate, Endo=m-Endo agar, FC=m=FC  agar,
17=rn-T7 agar, KF=m-KF agar,  MPNI=Total coliform MPN. MPN2=Fecal coliform MPN. MPN3=Fccal streptococci M.
tPerformance as measured by  Student Newman-Keuls  (SNK) Test.

-------
Table 16.   Media Performance in the  Assessment  of  Nontarget Bacteria.
1NHJJENTS
Average JRccoyery"'
Endo FC T7
Bayarn6n
Guaynabo
Villalba
Yauco
702
1923
2208
821
54
349
273
28
14
109
46
25
KF
63
87
174
287
MPN1
72
606
182
370
MPN2
20
260
316
67
MPN 3
45
136
228
236
tJCJUli?JLJBAn.Cl£!.T
lliph^C - 	 ~ 	 -^1 ow

Endo>FC=T7=KF=MPNl=MPN2=MPN3
Endo=FC=T7=KF=MPN 1 =MPN2=MPN3
Endo=FC=T7=KF=MPN 1=MPN2=M PN 3
Endo=KF=MPNl>FC=T7=MPN2=MPN3
EIT-LUENTS
Average Recovery*
Endo PC T7
Bayamon
Guaynabo
Villalba
Yauco
1
0
0
0
6

0
0
2
4

1
0
0
0
1 1 ; f ' i
KF
4
0
1
2

MPN1
2
3
1
1

MPN2
2
4
2
1

MPN3
2
1 4
3
2



Endo=FC=T7=KF=MPNl=MPN2=MPN3
Endo=FC=T7=KF=MPNl=MPN2=MPN3
Endo=FC=T7=KF=MPN 1 =MPN2=MPN3
Endo=FC=T7=KF=MPNl=MPN2=MPN'3

*Average Recovery =  Average  of the target colonies recovered on each plate, Endo=rn-Endo agar,  FG=m=FC agar
17=m-T7 agar. KF=m-KF agar.  MPNl=Total coliform MPN,  MPN2=Fecal coliform MPN,  MPN3=Fecal streptococci
fPerformance as measured by  Student Newman-Keuls (SNK) Test.

-------
                                                                                                              K -1
Table  17.  Specificity  and Selectivity  of  Viable  Count Methods in the Assessment  of  Influents.
PARAMETER

Presumptive Target
Verified Target
Presumptive nontarget
Verified nontarget
Specificity Indexes (%):
False Positive Errors
False Negative Errors
Selectivity Index (%)
n. d
ag
1 7
10
24
1 4

4 1
50
4 1
o m-FC
agar
37
26
22
1 6

30
1 9
63
m-T7 F
agar
20
15
1 0 4
8

25
1 2 3
<•, f. 3
MPN1

26
12
15
9

54
33
63
MPN2

30
1 9
28
20

37
29
52
MPN3

1 5
1 1
39
28

27
45
29
MPNl=Total  coliform MPN.  Mr
                                 2-.fViv..\J
I'N.'i
ptococx

-------
Table 18.  Specificity  and Selectivity of Viable Count Methods in  the Assesment of Effluents.
METHODS
PARAMETER

Presumptive target
Verfified target
Presumptive nontarget
Verified nontarget
Specificity Indexes (%)
False positive error
False negative error
Selectivity index (%)
m-Endo
agar
20
7
17
5

65
63
54
m-FC
agar
25
1 3
1 9
8

24
46
57
m-T7
agar
19
12
1 6
10

37
33
54
m-KF
agar
1 5
10
7
4

33
23
68
MPN1

21
6
7
3

7 1
40
75
MPN2

31
1 4
35
1 5

55
59
47
MPN3

1 5
9
4
2

40
1 8
79
MPNl=Total coliform MPN. MPN2=Feca! coliform MPN. MPN3=Fecal streptococci MPN.

-------
Table  19.   Bacterial Species Isolated  from Influent Waters.
Number of
Species
J^tlicjdthia £fiil

Acromonas hytfrophv
Unidentified*
Vibrio uiiirJLLCJLS
Pscudomopas sp.
Klebsiclla pJlcjajoiojUAS
CitrobacAcr, JTcjuidii
AjLhimiIO_b_a£lcjL JJDL.
Shi^clla dyscnieriac
Proteus vulgarls
Scrratia odorifcra

Times Isolated
73
27
La g
1 5
2
4
I 12
2 1
1
3
2
4
% Times
Isolated
42.7
15.8
5.3
8.8
1.2
2.3
7.0
12.3
0.6
1.8
1.2
2.3
% Recovery on Fecal ColifonnJSlcdk
m-FC
30.1
63.0
22.2
26.7
50.0
25.0
20.0
29.0
100.0
m-T7 MPN (EC Broth)
42.7 27.4
37.0 37.0
44.5 35.3
26.7 46.7
50.0
75.0
80.0
35.7
100.0
100.0
100.0
TOTAL - 170



Unidentifiable  by the  API-20E  System  for  the identification  of enterobacteriaceae  (Analytab)

-------
                                                                                                               K7
Table  20.   Bacterial Species
from Chlorinated Effluents.
Number of % Times
Species Times Isolated Isolated
liicJiiiitiiia toJU
KJcjjslcUa pjQfemicaojuac
Pscudomopas. sp.
Unidentified*
Citrobacicr frcundii
Entcrobactcr cloacae
22
3
6
3
2
5
61.1
8.3
16.7
8.3
5.6
13.9
% Recovery on Fecal Conform Media
m-FC m T7 MPN (EC Broth)
27.3 50.0 22.7
33.4 66.6
94.4 5.6
66.6 33.4
100.0
100.0
TOTAL = 41
•"Unidentifiable  by the  API-20E System for the  identification of enterobacteriaceae  (Analytab)

-------
                                                                      KH
         Figure  1.  Study Sites.

1 =  La Plata  Water Treatment  Plant  (Bayamdn)
2 =  Los Filtros  Treatment Plant (Guaynabo)
3 =  Villalba Urban System Plant (Villalba)
4 =  Ranchera Ward Treatment  Plant  (Yauco)

-------

-------
Figure  2,   Total Coliform  Densities  found in  Bayamon Influent Waters (September  24,  1987 to March  7.
          1988).

-------
     104
£
o
o
     10
       3 _
ID
LJL
O
     10
       2 .
 CO
 c
 0)
Q
     10' •:
      10'
                            MPNBi

                            m-Endo Bi
         0
468

    Time (weeks)
12
14

-------
Figure 3.   Total  Coliform Densities in Guaynabo  Influent  Waters (September  24.  1987 to March 7, 1988).

-------
 6      8     1
Time (weeks)
12
14

-------
Figure 4.   Total  Coliform  Densities in Yauco Influent Waters (September  24,  1987 lo March 7,  1988).

-------
   10'
o
   ioH
ID
LJL
O
CD
C

-------
Figure 5.  Total Coliform Densities in  Villalba Influent  Water (September  24,  1987 to March 7, 1988).

-------
    10'
o
o
10*:
o^


>»


c

Q
10J :
    10'
                           Time (weeks)

-------
Figure 6.   Fecal  Coliform Levels  in  Bayam6n Influent Waters (September  24,  1987 to March  7,  1988).

-------
10
                     Time (Weeks)

-------
                               . .   .   r                    --              24  1087  lo  March 7, 1988).
Figure 7.   Fecal  Coliform Densiues in Gu.

-------
    10
E
o
o
ID
LL.
O
    10
      3 J
>»  10
c

-------
Figure 8.  Fecal Coliform Densities  in  VillaJba Influent  Waters (September 24, 1987  to  March  7,  1988).

-------
£
o
o
1—

5
UL
O
10
        3 .
10
  2 _
c
0>
Q
      101 -:
      10°
              m-FC
       ----O—  m-T7

       	-a—  MPN
          0
           —r~

           2
8
10
12
14
                             Time (Weeks)

-------
Figure 9.   Fecal Coliform Densities  in Yauco Influent Waters  (September 24, 1987  to  March  1,  1988).

-------
10
   0
                    Time (Weeks)

-------
Figure 10.  Fec&l Streptococci Levels in Villalba  Effluent Waters  (September 24, 1987  io  March  7,  1988).

-------
   10
E
o
Y—
ID
LJL
O
CD
d
CD
Q
   10°
   10
     -1
      0
T-
2
-1	f
       6810

       Time (Weeks)
12
14

-------
Figure 11.  Influent AODC* Densities at all  Sites (September 24,  1987 to March  7.  1988).   'Acridine  Orange
          Direct Counts.

-------
  10'
(1)
0)10'
Q
  101
      0
                                                        Bayam6n
                                                        Guaynabo
                                                        Villalba
                                                        Yauco
6810
Time (Weeks)
12
14

-------
                                                                  99





                      APPENDIX  I
   ANOVA of Influent Water Quality  Parameters by Site.
  Parameter      F          df	P_
Temperature
pH
Turbidity
Total Chlorine
Free Chlorine
9.00
3.68
14.58
1.00
1.00
3
3
3
3
3
, 40
, 40
, 40
, 40
, 40
<0.001*
<0,05*
<0.001*
>0.05"
>0.05
'Significant

-------
                                                                 100





                      APPENDIX  II
    ANOVA of Effluent Water Quality Parameters  by  Site.






   Parameter	F	 df
Temperature
pH
Turbidity
Total chlorine
Free chlorine
24.25
2.28
4.24
16.63
27.04
3
3
3
3
3
, 40
, 40
, 40
, 40
, 40
<0.001*
>0.05
<0.05*
<0.001*
<0.001*
*Significant

-------
                                                              101





                    APPENDIX  III
     ANOVA of Influent Viable Count  Methods by  Site.






   Method	F	df
MF-Total coliforms
MF-Fecal coliforms (m-FC)
MF-Fecal coliforms (m-T7)
MF-Fecal streptococci
MPN-Total coliforms
MPN-Fecal coliforms
MPN-Fecal streptococci
5.08
4.75
1.56
2.82
11.39
15.68
16.57
3
3
3
3
3
3
3
, 40
, 40
, 40
, 40
, 40
, 40
, 40
<0.05*
<0,05*
>0.05
>0.05
«acoi*
«0.001*
«0.001*
* Significant

-------
                                                                  102





                      APPENDIX IV
      ANOVA  of Effluent Viable Count Results by Site.






   Method	F	df
MF-Total coliforms
MF-Fecal coliforms (m-FC)
MF-Fecal coliforms (m-T7)
MF-Fecal streptococci
MPN-Total coliforms
MPN- Fecal coliforms
MPN-Fecal streptococci
1.01
0.89
0.51
0.84
1.43
0.69
0.80
3
3
3
3
3
3
3
, 40
. 40
, 40
, 40
, 40
, 40
, 40
>0.05
>0.05
>0.05
>0.05
>0.05
>0.05
>0.05
*Significant

-------
                                                                   103





                          APPENDIX  V
                  ANOVA of Methods; Pooling Sites,






                            INFLUENTS

MF -
MF -
MF -
MF -
MPN
MPN
MPN
Method
m-endo agar
m-FC agar
m-T7 agar
m-KF agar
- total coliforms
- fecal coliforms
- fecal streptococci
F
6.870
7.220
2.840
0.003
52.75
66.100
46.600
df
3, 44
3, 44
3, 30
3, 44
3, 44
3, 44
3, 44
P
<0.001*
<0.0005*
>0.05
>0.25
<0.0005*
<0.0005*
<0.0005*
EFFLUENTS

MF -
MF -
MF -
\fF -
MPN
MPN
MPN
Method
m-endo agar
m-FC agar
m-T7 agar
m-KF agar
- total coliforms
- fecal coliforms
- fecal streptococci
F
0.057
1.370
0.930
0.545
21.690
11.310
2.760
df
3, 44
3, 44
3, 30
3, 44
3, 44
3, 44
3, 44
P
>0.25
>0.25
>0.25
>0,25
<0.0005*
<0.0005*
<0.05
'significant

-------