c/EPA
            United States
            Environmental Protection
            Agency
            Environmental Research
            Laboratory
            Narragansett Rl 02882
EPA-600/3-79-061
June 1979
            Research and Development
Metabolic
Responses of
Shallow Tropical
Benthic Microcosm
Communities to
Perturbation

-------
                RESEARCH REPORTING SERIES

Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology.  Elimination  of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:

      1.   Environmental  Health Effects Research
      2.   Environmental  Protection Technology
      3.   Ecological Research
      4.   Environmental  Monitoring
      5.   Socioeconomic Environmental Studies
      6.   Scientific and Technical  Assessment Reports (STAR)
      7.   Interagency  Energy-Environment Research and Development
      8.   "Special" Reports
      9.   Miscellaneous Reports

This report has been assigned to the ECOLOGICAL RESEARCH series. This series
describes research on  the effects of pollution on humans, plant and animal spe-
cies, and materials. Problems are assessed  for their long- and short-term influ-
ences. Investigations include formation, transport, and pathway studies to deter-
mine the fate of pollutants and their effects. This work provides the technical basis
for setting standards to minimize undesirable changes in living organisms in the
aquatic, terrestrial, and atmospheric environments.
  n   ~Qi8-av?lS>lf,!°thepublic throu9h the NationalTechnical Informa-
tion Service, Springfield, Virginia  22161.

-------
                                              EPA-600/3-79-061
                                              June 1979
   METABOLIC RESPONSES OF SHALLOW TROPICAL
BENTHIC MICROCOSM COMMUNITIES TO PERTURBATION
                      by
                 S.  V.  Smith
                 P.  L.  Jokiel
                  G.  S.  Key
                E. B,  Guinther
      Hawaii Institute  of Marine Biology
            Kaneohe, Hawaii 96744
            Grant No.:  R800906
              Project Officer

              Kenneth T. Perez
     Environmental Research Laboratory
              South Ferry Road
     Narragansett, Rhode Island 02882
     ENVIRONMENTAL RESEARCH LABORATORY
     OFFICE OF RESEARCH AND DEVELOPMENT
   U. S. ENVIRONMENTAL PROTECTION AGENCY
     NARRAGANSETT, RHODE ISLAND 02882

-------
                                 DISCLAIMER
     This report has been reviewed by the Environmental Research Laboratory,
Narragansett, U. S. Environmental Protection Agency, and approved for
publication.  Approval does not signify that the contents necessarily reflect
the views and policies of the U. S. Environmental Protection Agency, nor does
mention of trade names or commercial products constitute endorsement or
recommendation for use.
                                      ii

-------
                                  FOREWORD
     The Environmental Research Laboratory of the U. S. Environmental
Protection Agency is located on the shore of Narragansett Bay, Rhode Island.
In order to protect marine resources, the.laboratory is charged with
providing a scientifically sound basis for Agency decisions on the environ-
mental safety of various uses of marine systems.  This requires research on
the tolerance of marine organisms and their life stages, as well as eco-
systems, to many forms of pollution stress.  In addition, a knowledge of
pollutant transport and fate is needed.

     The report that follows describes the use of flow-through aquaria for
establishing, maintaining and monitoring shallow tropical benthic communities.
Such studies are a logical intermediate point between laboratory bioassays
and field surveys.  The project is also intermediary between an earlier
analysis of coral response to thermal stress and a present analysis of the
responses of an entire ecosystem to the termination of sewage stress.
                                       Eric D. Schneider
                                       Director
                                       ERL, Narragansett
                                     iii

-------
                                  ABSTRACT
     Benthos communities simulating various aspects of coral reefs were
established in 600-liter microcosm tanks.   These communities were then
subjected to various environmental perturbations, including altered light
regime, altered substratum type, salinity depression, elevated nutrient
level, and biological manipulation.  The metabolic responses of the
community to these perturbations were monitored, primarily by analysis
of dissolved oxygen flux.  Light, substratum type, and nutrient levels
are resources which limit community metabolism.  From 35 to 22 °/oo,
metabolism is not sensitive to salinity.  Salinities below 22 °/oo kill
most test organisms.  Metabolism is sensitive to biological manipulation.
                                      iv

-------
                               CONTENTS
Foreword	
Abstract	     iv
Figures	  .     vi
Tables	•  •  •  •	   viii
Acknowledgments	     ix

     1.   Introduction   .......  	      1
     2.   Conclusions .....  	      6
     3.   Recommendations	      8
     4.   Utilization ......'	      9
              A.  Description of microcosms used in
                  this  study	      9
              B.  Manipulation of the microcosm environment   .  .      9
              C.  Methods of structural manipulation and
                  analysis  in microcosms	     11
              D.  Methods of functional analysis in
                  microcosms .....  	 .......     11
              E.  Flushing characteristics of microcosms
                  and natural ecosystems	     13
              F.  Gas exchange in HIMB microcosms	     16
              G.  Relationship of A02 to ACOa	     16
              H.  Replication and reliability of
                  metabolic measurements  .... 	     17
     5.  Microcosm Community Metabolism   ............     19
              A.  Light	     19
              B..  Substratum	     29
              C.  Nutrients  ........ 	     33
              D.  Salinity	     36
              E.  Biological manipulation	     39
              F.  Conclusions  ......... 	  .  .     42

References	•	     44
Appendix:   Design, Construction, and Operation of Shallow
     Tropical Benthos Microcosm Facilities ...........     47

-------
                                FIGURES


Number                                                              Page

   1   Configuration of the HIMB microcosm facility	    10

   2   Return of altered salinity to ambient as a.function
         of tank flushing	    15

   3   A.  Curve of ambient solar radiation over 24 hours ....    20

       B.  Dlel response of oxygen flux with and without
             fouling communities  	    20

   4   Solar radiation versus oxygen flux for four successive
         measurement periods in the same tank . , .	    21

   5   Comparison of oxygen fluxes among replicate treatments
         under identical light regimes	"..	    24

   6   Solar radiation versus oxygen flux treatments  	    25

   7   Comparison of replicate tank oxygen fluxes .	%    27

   8   Solar radiation versus oxygen flux of benthos
         communities developed on various substrata .......    30

   9   Direct comparison of oxygen flux on rubble versus
         sand and mud substrata	    32

  10   Solar radiation versus oxygen flux for "low nutrient"
         (bay water) and "high nutrient" (bay+well water)
         treatments .	    34

  11   Direct comparison of "low nutrient" (bay water) versus
         "high nutrient" (bay-Hwell water) treatments	    35

  12   Solar radiation versus oxygen flux under various
         salinity regimes .	    38

  13   Solar radiation versus oxygen flux in tanks with and
         without herbivorous fishes  	    40

  A-l Map showing  locations of HIMB and NUC microcosm facilities    48
                                    vi

-------
A-2  A.  Arrangement of standpipes in Inlet headbox ......   52

     B.  Arrangement of standpipes in outlet headbox  	   52

A-3  Schematic diagram depicting the HIMB water sampling
       system	   54

A-4  Enlargement of area of water sampling system (see
       Figure A-3) which measures oxygen, salinity
       and temperature  	  .........   55
                                vi±;

-------
                                 TABLES
Number                                                              Pa8e

   1  An example of typical community succession 	   ^2

   2  Regression equations for data presented in Figure 4   ....   22

   3  Analysis of covariance for data in Table 2 and Figure 4  . .   23

   4  Analysis of covariance for data in Figure 5	   25

   5  Analysis of covariance for data in Figure 6	   26

   6  Analysis of covariance for data in Figure 8	   31

   7  Analysis of covarianca for data in Figure 6A, B and
        Figure 8A	   31

   8  Comparison of typical dissolved inorganic nutrient levels
        in bay water drawn through microcosm facility with  sea
        water well	   33

   9  Analysis of covariance foe data in Figure 10 ........   35

  10  Organisms introduced into fouling communities
        for salinity stress experiments  	 .....   36

  11  Chronology of salinity stress experiments	   37

  12  Analysis of covariance for data in Figure 12 	  ...   39

  13  Analysis of covariance for data in Figure 13	   41

  14  Analysis of covariance, no fish versus fish, rubble   ....   41

  15  Analysis of covariance, reef rubble versus
        echinoids versus  fish   ......  	  ......   42
                                    vlii

-------
                              ACKNOWLEDGMENTS


     We are grateful to our many colleagues at the Hawaii Institute of Marine
Biology, Kaneohe, Hawaii, and at the Naval Undersea Center (now the Naval
Ocean Systems Center), Kailua, Hawaii, for their considerable assistance on
various aspects of the study.  Dr. John E. Bardach served as principal
investigator during the course of the study, and Dr. Sidney J. Townsley did
so during its initial year; we thank them both.  We owe a particular debt of
gratitude to the many students who worked long, and largely underpaid, hours
to insure the success of the program.
                                     ix

-------
                                  SECTION I

                                 INTRODUCTION

      Biological communities in natural settings are the product of a complex
 and largely unknown history of physical-chemical stimuli and organism re-
 sponses and interactions.  Attempts to explain the composition and function
 of communities and organisms in natural settings often do not arrive at a
 quantitatively satisfactory understanding of how the communities might change
 in response to perturbation.  Biologists have frequently turned to laboratory
 or aquarium studies of single organisms or populations to avoid the complex
 and uncertain history of natural communities.  Attempts to build a community
 analysis from such observations are ordinarily frustrated by the lack of
 realism in physical,  chemical,  and biological characteristics within the
 aquaria.  The logical compromise is to maintain several species and trophic
 levels of organisms as an ecologically functional unit, and expose this unit
 to realistic and controllable physical-chemical conditions.  The term
 "microcosm" has been applied to various artificially maintained biological
 assemblages that function as largely self-contained functional units.   Beyers
 (1963)  lists the following terms as having been applied to artificial aquatic
 ecosystems:  microcosm,  aquarium microcosm,  carboy microcosm, microecosystem,
 experimental ecosystem,  and laboratory-scale model.  Other terms have appeared
 in the  ecological literature since that time,  including "synthetic microcosm"
 (Nixon,  1969),  "gnotobiotic ecosystems" (Taub,  1969),  and "artificial open
 systems" (Confer,  1972).   The term "microecosystem," while perhaps more de-
 scriptive than "microcosm," is awkward , and  not  entirely satisfactory.   Thus,
 the term "microcosm"  will be used  throughout this  report.

     Using  microcosms to  study  effects  of,environmental change in aquatic
 environments  represents a logical  advancement  in  the evolution of pollution
 research.   Early  aquatic  pollution research  was mainly  concerned with  such
 public health problems as the spread of disease through contamination  of
 drinking water.   During the 1930's  and  1940's  it became apparent that  new
 tools were  needed  to  understand  the impact of pollution on  aquatic systems.
 Efforts led to the  development of the toxicity bioassay, which uses a single
 organism, (usually  a fish)  to "model"  the response of an ecosystem to pollu-
 tion stress, ,,,During  the  1950' s and I9601s the toxicity.bioassay  became an
 increasingly sophisticated  technique.

     The earliest  tests were short  (24-hour) static  tests.  Further work
 showed that the toxicity bioassay model became more predictive of  true bio-
 logical impact as  exposure  time,was  increased  (to 48 hours, 96 hours, and  even
 longer) and if continuous flow was  used.  An extensive body of published in-
 formation .on.proportional diluters  andmethods of analyzing mortality data
 developed.  The model was refined by including considerations of synergistic
effects such as natural light rhythm, temperature, oxygen tension, and,

-------
reproductive state; use of the most sensitive organism as an indicator further
refined the assays.  Another refinement was use of the most sensitive life
stage of the organism as the bioassay model.   The model was further refined
by using measures of metabolism (e.g. fish respiration), rather than mortality,
as the index of stress.  Other advances led to the inclusion of a full range
of test organisms (molluscs, crustaceans, etc.), in addition to fish, in the
14th edition of Standard Methods for the Examination of Water and Waste Water
(Anonymous, 1976).

     During the last decade, it has become increasingly evident that even
more sophisticated and complex analytical techniques are needed to predict
ecological effects of pollution.  The study of complex multi-compartment
living systems and their physical environments plays an increasing role in
pollution research.  It was inevitable that researchers would develop more
complex simulations of complex natural systems.  These simulations were
microcosms, a logical extension from classical single-organism bioassays to
a multi-component analysis.  The microcosm includes all the features of the
bioassay and differs only in complexity.  The methodology, philosophy, and
application of the bioassay that developed over the years  led  directly to
microcosms.  Use of metabolic variables, natural light regimes, prolonged
sublethal stress, and continuous-flow apparatus originated with classical
bioassay technique.

     The microcosm provides an interface for interaction among various
scientific disciplines.  At the same time that pollution researchers were
moving from single organisms towards organism interactions, ecologists were
also in a period of transition.  Classical ecology began with studies of
natural systems in the field.  Difficulties with interpreting and manipulat-
ing natural communities led to simplifications in the laboratory.

     The microcosm is, at once, a complicated version of the pollution
researchers' original bioassays and a simplification of the ecologists'
traditional field observations.  The microcosm approach will not replace
existing techniques but will be used to study questions not amenable  to
present investigative methods.  In order to include the microcosm technique
in ongoing programs, one must consider potential uses and  alternative
techniques.

     In many instances, the microcosm is clearly the method of choice.  Popu-
lation or community level trophic interactions are not predictable from data
gathered on single organisms.  Size and complexity of natural ecosystems
ordinarily make in situ manipulation impractical.  The microcosm represents
an experimental approach which has been used successfully  to isolate  and
analyze processes in aquatic ecosystems.  Examples of this type of study in-
clude predator-prey interactions  (Hall et al., 1971), pelagic food-chain
dynamics  (Mullin  and Evans, 1974), diurnal metabolic patterns  (Beyers, 1963),
patterns of autotrophic succession  (Cooke, 1967), and the  recycling  of
sewage  effluent materials  through marine food  chains  (Ryther et al.t  1972).

     Microcosms may be constructed'and  scaled  according to differing criteria.
One  criterion  might be to  attempt to simulate  an entire ecosystem in terms  of
 the  concentration and  relative abundance of major abiotic  and biotic

-------
 components.   Our  criterion has been  to simulate relatively restricted com-
 munities,  or  parts  thereof.   Ecosystem-level simulations  can be achieved "by
 linking  (e.g.  via water  flow) community-level microcosms.

     Microcosms represent the logical union of laboratory and field investi-
 gative technique.   Hence, microcosm  experiments utilize various types of
 laboratory techniques and concepts developed for use on single species  (e.g.
 mortality  measures, metabolic techniques, growth measures, etc.)> combined
 with standard  field sampling  techniques and concepts (e.g. fouling panels,
 subsampling the community for epifauna, infauna, plankton and other compo-
 nents, sediment analysis, community  photosynthesis and respiration, etc.).

     The various  techniques of microcosm analysis fall into two broad
 categories:   functional  analysis and structural analysis.  Odum (1962) de-
 fines biological  "structure" of an ecosystem to be the composition of the
 biological community including species, numbers, biomass, life history, and
 distribution in space of populations.  That is, biological structure is the
 nature of  the  biota in an ecosystem  at an instant in time.  Odum defines
 "function" to  be  the rates of material, energy, and information flow through
 the ecosystem.

     Function  is  often amenable to direct total-system monitoring; such
 integrative monitoring reduces the need for the laborious and inherently im-
 precise summation of structural components to understand the system.  For
 example, many  organisms  contribute to the biomass of a community; carbon flux,
 which is a time-differential function of biomass, can be directly measured in
 the water  column  and can be used to  assess the change in biomass with time.
 The assessment is properly weighted  and integrated according to the contri-
 bution of  each component to ecosystem changes, whereas biomass need not have
 any direct and/or constant relationship to energy flow or material flux.
 Moreover,  non-destructive methods of measurement can be employed with
 functional assessment.

     Because microcosm technique invites the simultaneous use of numerous
 methods and involves very complex multi-compartment systems, it is vital not
 to lose sight  of  this unifying relationship between structure and function.
 Ultimately the goal of environmental research is to predict the exact nature
 of the structural/functional relationship for ecosystems influenced by human
 activity.

     Inclusion of microcosm facilities in a;well-planned environmental pro-
 gram provides  many  research advantages, but the technique must.also be evalu-
 ated in terms  of .time and cost.  Advantages and disadvantages which will
 ultimately influence utilization of  continuous-flow microcosms can be sum-v
 marized as follows:

Advantages

     1.  Microcosms are the most suitable method available for studying cer-
 tain community-level processes; the  size and complexity of natural eco-
 systems make them impractical to manipulate experimentally.  Experiments
 based on single organisms do not predict community responses.  The microcosm

-------
consists of major biological components of natural communities, and there-
fore should be a better tool for predicting ecosystem responses to perturba-
tion than are single organisms.

     2.  Various structural and functional responses can be observed simul-
taneously.  Growth, reproductive state, recruitment, mortality, individual
component metabolism, material and energy pathways can be identified and
measured.  Sensitive indicator species for a given stress can be quickly
identified.

     3.  Microcosms are extensions of the natural world.  Hormal diel varia-
tions in ambient conditions can be duplicated.  Natural foods and recruit-
ment are provided.  Normal seasonal variations are followed, and natural
serai development is allowed.

     4.  Wastes, biogenic toxins, hormones, and pheromones can be kept at
near-natural concentrations.

     5.  Microcosms represent a middle stage between controlled laboratory
experiments and in situ field observations.

     6.  Microcosms offer research flexibility.  Units can be connected in
series; thus progressive de-toxification or stripping processes may be
studied.  Conversely, a single unit may be simultaneously used for a number
of non-destructive studies.  Microcosms may be used as a sorting technique
for the identification of potential bio-indicators.

     7.  Various perturbants can be introduced into replicate microcosm
communities or communities with specified differences under controlled con-
ditions.  Sub-lethal stress levels may be used, because long-term experiments
are feasible; such experiments also lend themselves to chain-response
studies in complex communities.

     8.  Electrochemical probes and automatic recording systems can be added
to the microcosms, facilitating continuous observation.  Large amounts of
data gathered in this manner can be processed automatically.

     9.  Because microcosms can be more precisely controlled and monitored
than the natural systems they simulate, the microcosms are amenable to the
validation of mathematical ecosystem models.  Accurate mathematical descrip-
tions of natural systems are a major objective of the  environmental sciences,
because only with high-speed simulation of ecological pressures can a proper
evaluation be made of the likely long-term environmental effects of manage-
ment decisions.

Disadvantages

 i   .,-h  T?e COSt  °f a comPlex microcosm facility is high,  and considerable
 lead time  is required to build such a  facility.

-------
      2.  Nearly continuous pump operation is required in flow-through micro-
 cosms; there are high maintenance costs, as well as considerable experimental
 risk of breakdown.

      3.  Biological recruitment into the system may be largely uncontrolled,
 or controlled with difficulty.

      4.  There are constraints on the manipulation of certain variables
 (e.g. salinities higher than ambient are possible but difficult) and also on
 the size of the microcosm.  Although size limits have not been thoroughly
 investigated, not all components of a given community (large carnivores, for
 example)  can be effectively accommodated.

      Within this context of potential and realized utility of aquatic micro-
 cosms,  this report describes the design and utilization of microcosms
 appropriate for shallow tropical benthos communities.   Because these com-
 munities  may be complex, heterogeneous, fragile, and metabolically active
 (e.g.  coral reefs), we have been forced to consider and overcome numerous
 specific  operational problems.   Many design characteristics of the resultant
 product,  as well as many of the scientific findings, can be transferred to
 other situations.   We refer,  in particular,  to  the report by Henderson et a1<
 (1976)  comparing the Hawaii Institute of Marine Biology microcosm facility
 with  a  companion facility designed by the Naval Undersea Center.

      The  focus  of  this study  has been the design and construction of an
 experimental microcosm facility for  the culture of integral benthic  communi-
 ties, under conditions which  are more representative of  natural  situations
 than  simple laboratory aquaria,  and  at  the same time, are easier to  control,
 manipulate,  and monitor  than  the complex communities of  real-world coral
 reefs.

     Design,  construction,  and  utilization of the  facility  were  interrupted
 by  funding  breaks between  each  fiscal year of the program.   These  interrup-
 tions complicated the  maintenance of  personnel,  continuation of  experiments,
 and execution of a well-designed experimental chronology.   As  a  result,  the
 specific environmental lessons  learned  from our  efforts  are  less than  they
might have been.  Nevertheless, we have  extracted a  series  of  specific
accomplishments which demonstrate the utility of the microcosm facility and
provide insight into shallow tropical benthos ecosystems.

     There have been numerous spinoffs of this program.  COa has been
investigated as a metabolically useful variable, primarily in situations
other than the microcosms.  We have participated with other groups in field
studies at Canton, Christmas, Enewetak, and Fanning Atolls, in the tropical
Pacific Ocean.  We have worked with the Naval Undersea Center in the design
and utilization of a second microcosm facility.
                                      5

-------
                                SECTION II

                                CONCLUSIONS
     1.  General conceptual designs are presented for a microcosm facility
appropriate for simulating shallow tropical benthos communities.   These
designs present major considerations for the effective operation of such a
facility.

     2.  The microcosm facility proves to be a useful tool for the study of
benthic communities over periods of several months.  The fouling communities
which develop in the microcosms can be complex, and organisms deliberately
introduced into the facility function normally.

     3.  Microcosm tanks built according to our design can be characterized
by a physical flushing model of total mixing; for these microcosms, 02 pro-
duction has proven to be a useful metabolic measure.  Gas exchange can be
ignored.

     4.  Shallow benthos communities which have been set up in the tanks are
demonstrably light-limited.  The solar conversion efficiency of these com-
munities is high, so naturally or artificially turbid water is likely to be
a significant detriment to the rapid productivity characteristic of reef
communities.

     5.  Substratum is a second resource which limits the metabolic activity
of microcosm communities.  Substratum type is often altered by artificial
activity.

     6.  The microcosm communities are also limited by nutrient availability,
even though the nutrient loading at our primary microcosm site is high.
Again, nutrient loading is a variable often altered by human impingement on
reef ecosystems.

     7.  At least three variables  (light, substratum, nutrients) are  resources
which  may  simultaneously limit reef production.  Thus, heterogeneous benthos
communities, even as  simplified in the microcosm situation, do not conform to
Liebig's Law of the Minimum.

     8.  Salinity down to  22   /oo  does not have  a  demonstrably deleterious
effect on  reef ecosystems.  The damage  done  to reefs  by  stream runoff must
therefore  be attributed  either to  lethal  effects due  to  extreme  salinity
depression or  to associated variables  such as  siltation  or  toxicants.

-------
     9.  Within broad limits, organic carbon production by benthic communi-
ties with varying community structure is similar.   If grazing pressure is
altered, metabolic response is responsive to quantity (but not particularly
to quality) of altered grazing pressure.

-------
                                SECTION III

                              RECOMMENDATIONS
     1.  Microcosms should be established and used as logical extensions of
standard bioassay techniques in order to gain a more adequate understanding
of how ecosystems respond to stress.

     2.  The results reported here are restricted to community oxygen metab-
olism.  Examination of C02 and nutrient fluxes should be undertaken, although
oxygen metabolism is by far the easiest variable to automate.  Community
structure in the microcosms can also be examined, although variables related
to structure are almost as "noisy" in the microcosms as in natural situations.

     3.  The oxygen metabolism data are examined by analysis of covariance
with the regression analysis being a convenient method to screen out the    '
primary community response to variable  (and largely uncontrolled) light
levels .

     4.  Extended, systematic experiments should be conducted to validate  and
refine  the results of preliminary experiments reported in this document
Light,  substratum, and nutrients are all demonstrably important to reef
metabolism and  should be  investigated further.  Salinity effects amjear
surprisingly minor; this  point should also be examined in more detail   Bio-
logical manipulation is also an important control of community metabolism


-------
                                   SECTION IV

                                  UTILIZATION


  A.    Description  of Microcosms  Used  in  this Study

       A  detailed description  involving the design,  construction and operation
  of  the  HIMB-NUC microcosm  facilities is  in  the Appendix.  The following
  section describes briefly  specific experimental aspects of the microcosm
  units.

       The microcosm  tanks are 117 cm square  by 46 cm deep fiberglass containers
  (Figure 1).  A continuous  gravity-regulated jet (normally 10 liter min-1)
  of  seawater enters  the microcosm through a  port in the center of the floor
  and leaves the system through a drain in the floor corner.  This arrangement
  provides mixing of  the water without excessive agitation, local turbulence
  or bubble formation.  Violent entrapment of air alters dissolved oxygen flux
  used  to calculate metabolism, so such air entrapment must be avoided if
  community metabolism is to be measured.   Mixing characteristics of the micro-
  cosms are discussed elsewhere in this section.  The microcosm tank is a well-
 mixed, continuous-flow, 630-liter reaction vessel with a mean water residence
  time of approximately 1 hour.  Inlet and outlet water composition can be
 sampled automatically for some variables (dissolved oxygen,  pH,  temperature)
 or manually for others (inorganic nutrients, particulate load,  plankton,
 alkalinity,  etc.).

      Alteration of chemical,  physical, or biological characteristics  of the
 influent water is  carried out before it  enters the  microcosms.   The particu-
 lar  tanks used in  this study  are a manageable size  suitable  for  simulating
 the  community  structure of  shallow tropical  benthic systems.

 B.   Manipulation of the Microcosm Environment

     The effort to design adequate microcosm facilities was directed at pro-
 ducing various living multi-compartment representations of reef communities,
 and  the  means of manipulating the environment within these communities.
 During this development program, a variety of factors have been manipulated:

     1.  Chemical factors—The initial manipulation of water chemistry was
 aimed  at dissolved nutrient elevation.  During the studies reported here,
 nutrient elevation was accomplished by mixing relatively low-nutrient water
 from Kaneohe Bay with high-nutrient water from a seawater well.   Subsequent
 nutrient-loading experiments have been accomplished by using a precision
 peristaltic pump to add concentrated nutrient salt solutions at 1 to 100 ml
min-1  to the predominantly low-nutrient flow (10 liters min"1).   Copper

-------
                                                            SOK7JU5
                                                          LAB
                                           IKLCT UATIit LIKES    •'    0?riCC
                                 SAtr VttL fJHt
                          U.S.C.CS- TISE
                           &1ATION
                                                                ST/VATIS WtXINS_
                                                                K A KI F 01 Q^r*"*?
                                   ^*.*^   ^---f

                    «•-• -^r»e*w»ur-=!iJ'*«-DiH« TA^K_-!  - _.  -_ ^->  --_•  -•
                    ••~~-"*~^. -^tl«£i '	—=r"*5:. -     OAAIM -     __  ^-^Sv.
                          IMTAKC
             Figure 1.   Configuration of the HIMB microcosm facility.
additions at the NUC facility have  similarly been effected by the addition
of concentrated salt solution.   In  both the reagent additions and the bay
plus well mixtures, the microcosm head boxes serve as mixing chambers.

     Salinity alteration was accomplished by adding municipal fresh water
from a large (10 m3) holding tank maintained at a constant head by a vertical
standpipe and an excess of  freshwater input.  Water was gravity-fed at a
constant flow rate  from this tank to the microcosm inlet boxes.  This pro-
cedure prevented possible variation in delivery rate from variation in line
pressure.

     Dissolved oxygen  levels were levered by bubbling nitrogen through the
seawater headbox.   The degree  of 02 lowering may be adjusted by varying  the
rate at which the N2 is bubbled through the inlet head box.

     2.  Physical  factors—Light is one of the most important physical vari-
ables, because it represents  the major energy source for plant communities.
The light intensity, spectral  quality, and photoperiod encountered  on tropi-
cal reefs cannot be easily  duplicated in a laboratory.  Fortunately,  there is
                                       10

-------
no  reason to  duplicate  such conditions,  because microcosms  can  usually be
located outdoors  in full  natural  radiation.   Reduction  of light below ambient
levels  (such  as might occur with  high water  turbidity)  can  be simulated by
placing various layers  of different  neutral-density  screening over  the micro-
cosms  to reduce light.  By use  of multiple screens of various densities, we
have worked at light levels between  2 and 70 percent of full sunlight.

     Temperature  is altered by  heater-chiller units  as  described by Jokiel
et  al.  (manuscript).  Substrata of various types are added  as required.  Mud,
sand, rubble,and  hard substrata have been utilized in this  program.  Water
motion  is increased by positioning  motor driven agitators of  various
horsepower on the microcosms.

C.   Methods  of Structural Manipulation  and  Analysis in Microcosms

     The microcosm  communities  used  in this  study are complex in biological
structure.  Several hundred species  of organisms commonly establish them-
selves,  entering  the microcosms as larvae in the seawater.  These communities
are rich, complex,  and responsive to changes in microcosm environment (e.g.
alteration of substrata,  water  motion, nutrient loading, light  regime,
presence of large grazing fish, etc.).   As in the case  of natural communities,
an  orderly community succession occurs (Table 1).  An initial bloom of algae
is  followed by settlement of rapidly growing herbivores  such as  the sea hare
Stylocheilus  longieauda,  which  crop  back the algae.  Other  organisms settle;
in  time  (usually  several  months)  carnivores  and omnivores settle, and a
community which will remain relatively stable through time  develops.

     Transplanting  intact communities from the field can shorten the period or
alter the characteristics of succession  and  growth.  Transplanted communities
are less  dependent  than the fouling  communities on larval recruitment, and
stabilize to ambient tank conditions rapidly.  Substrata with associated
organisms are removed from the  field and transported to microcosms.  The
microcosm community  can also be structured by adding fishes, corals, or other
fully grown organisms that  otherwise would enter randomly and grow  towards
adult size during the course of the  experiments.   Some  large benthic organisms
of a coral reef have lifespans  of months to years,  and  community succession
in reefs  is both  complex  and slow.   It is therefore impractical to wait for
larval recruitment  to produce a truly mature  community  in the microcosms.

     As  experiments are conducted, routine structural subsampling can be
carried out,  usually using many of the same  techniques  that are employed in
field programs.   Various  subsampling routines have been employed, including
subsampling the substrata, scraping  small sections  of the wall,  or utilizing
fouling panels.   Photography is a satisfactory record for many analyses.
Various methods of statistical analysis of community structure may be applied.
Fishes, coral, and other  large biota can be removed from the tanks,  weighed
individually,  and then returned to the tanks.

D.   Methods  of Functional Analysis in Microcosms

     The  results presented in this report deal primarily with community
function of the microcosms.  Measurement of microcosm community  function

                                      11

-------
              TABLE 1.   AN EXAMPLE OF TYPICAL COMMUNITY SUCCESSION

                          Chronology of first  appearance of  dominant
                          biota  in microcosms  at the  NUC facility
                          (Spring 1975),  starting with "sterile"
                          microcosms (from Henderson  et a£.,  1976).
       Weeks of
        Flow-
       Through
               Oiideriant              Annelids            Echinoderms
   Algae                  Crustaceans          Molluscs
         1+


         2+

         3+

         4+

         5+
         7+
         8+
         9+
        10+
        11+

        12+
   diatoms
 cyanophytes
 filaments and
 tufts of algae

  green tufts
calcareous algae
                          amphipods
               brownish algae tuftj
 Ectocarput
   Valonia
  Laurenda
                             Aiptasia pulchella
                          portunjds

                      Pennon planissimum
                                       StylocheOus hngicauda
  Ctrithium nesioticum

     bubble shells
Cheilidonura hirundintjina
                Synaptidae sp.
   other nudibranchs
    Aplysia parvula
    Aptysia Juliana
 Dotabrifera dolebrifero
                                        DolebeUa euricularia
                                             vermetids
                                  ffydroides sp.
                           alpheids    ipiiorbids
Padina
13+ Hydroclothrus
Colpomenia
14+
• f m
15+
16+
17+
18+
19+



ffippoiysmaltt kukenthalt

Gnethophyllum faciolatum







Strombus maculatus
Cerithiumtinenfis

Cypraea caputterpentis
Echinothrix sp.
Srichopus horrent
Pinna muricata
involves  the measurement  of net  flux of various biologically active materials
(e.g.  oxygen, carbon dioxide, nitrogen compounds, and phosphate)  through the
system.   Oxygen is used most often because it is easily monitored,  can  be
recorded  continuously with a polaragraphic cell, and is a  direct  measure of
                                           12

-------
  net photosynthesis and respiration.  There is ordinarily relatively little
  variation in the inlet oxygen concentration, so the flux calculations are
  straightforward (see discussion below).  Such measurements are by no means
  restricted to oxygen.

       Flux of nitrogen  or phosphorus is determined by measuring outlet and
  inlet concentrations of the various biologically active forms  of  these
  elements.   Large variations in the inlet  nutrient levels make  nutrient fluxes
  difficult  to calculate without large numbers of analyses.   Measurement of
  total carbon flux can  be determined using pH,  alkalinity,  dissolved organic
  carbon and particulate organic carbon measurements.   Net calcification within
  the  community can be determined from alkalinity changes alone.  Smith and
  Key  (1975)  summarize the use of C02 as a  metabolic record  in marine eco-
  systems.   Flux of toxicants such as introduced heavy  metals, or chlorinated
  hydrocarbons,  as  well  as their effect on  community metabolism, can  be
  measured.   After  a period of addition,  rate  of de-toxification and  recovery
  can be  similarly  determined.

       If inlet  concentration,  outlet concentration, and  flow rate are  known,,
  the total flux of  the material for  a steady  state  microcosm would be
  calculated  as  (outlet concentration -  inlet  concentration) times .the  flow
  rate.  Steady  state  conditions  are  not usually obtained, because most
 metabolic processes  follow a  diurnal rhythm, and because of short-term
 variation.  Hence, mixing characteristics of the microcosm must also be
 considered.

 E.   Flushing Characteristics of Microcosms and .Natural Ecosystems

      In order to measure community metabolism,one must .understand the flush-
 ing characteristics of  the ecosystems in order to apply appropriate calcula-
 tions to the conservative (advective or diffusive) and nonconservative
 (-in situ uptake or release)  fluxes  of metabolically relevant variables.  One
 may imagine three end members of flushing  models on a triangular diagram:
 non-mixing  stream flow,  complete-mixing with continual exchange, and pulsed
 exchange.   Real world conditions are ordinarily intermediate among these end
 members,  but many practical  situations cluster  near one of  these conceptual
 extremes.   It is often  possible to  take advantage of  a natural  or  artificial
 tracer which is strictly conservative in order  to describe  the  appropriate
 flushing model.  Smith  and Pesret (1974) and  Smith and Jokiel  (1976) have
 applied such a technique with salinity in  coral reef  ecosystems  (see also
 Smith,  1974).

     It would appear  that the complete-mixing,  continual-exchange model most
 appropriately-describes  the  characteristics of  the  microcosms.  If.this
 hypothesis  is correct,  then  any instantaneous alteration.of a totally
 conservative property, of  incoming waters should be  characterized by  an
 exponential  decay,curve  from the pre-afteration state  to  the post-alteration
 state.  The  decay  constant should be the inverse residence  time.of water in
 the tank (i.e.  the  flow rate  of water  through the tank divided by tank
volume).
                                     13

-------
     Salinity changes during a salinity-alteration experiment demonstrate the
validity of this complete-mixing model for our tanks.  The salinity of the
inlet box was abruptly dropped 10 °/oo below ambient (from about 35 °/oo to
25 °/oo) by an admixture of fresh water, and the salinity -in the tanks was
monitored for four hours.  Figure 2 demonstrates that,  after an initial lag
(related, perhaps, to density stratification) the salinity drop was well
approximated by a negative exponential (mixing) curve.   The exponential slopes
yielded water residence times ranging from 53 to 64 minutes; these residence
times agree well with the 60 minute residence time being approximated during
that experiment by the tank flow rate.

     The general validity of the complete-mixing model for the microcosms has
been established with conservative properties.  We can now turn to nonconser-
vative properties and superimpose these on the model.  For a known residence
time, we can now develop the appropriate equations.  Let the subscripts 0 and
t denote samples at times 0 and t minutes later; T is the residence time of
water in the tank.  X is the inlet concentration  and Y is the outlet con-
centration; "a" is a subscript denoting the concentration change of Y attribu-
table to advective flux; "r" denotes the concentration change  attributable to
internal reactions. V is the tank volume; Z is the tank depth.  F is the flow
rate of water through the tank.

     The residence time may be calculated:

          T- V/F                                                         (1)

The advective concentration change  at time t may be approximated by:

          Ya =  [
-------
z
_l
<


<3
     10.0
     0.1
	• TANK 7 - INLET C



	O TANKS- INLET C




	O TANK 9- INLET C
T Cmin)   r2


  53   0.98



  61    0.99



  64    1.00
             DAY 195  SALINITY STABILIZATION
                  I          2          3         4


                           TIME (HOURS)





       Figure 2.   Return of altered salinity to ambient as

                  a function of  tank flushing.
                             15

-------
F,   Gas Exchange in. HIMB Microcosms

     Measurements of metabolic rates according to the general flux model
developed in the previous section implicitly or explicitly assume that advec-
tion of water and state changes within the water are the only two processes
of concern.  That is not quite true; there can be gas exchange of materials
across the air-water interface.  Oxygen is a commonly (including in our
studies) used metabolic indicator, and oxygen exchange across the air-water
interface can be a significant flux term (e.g. Kanwisher, 1963; Kinsey, 1973).

     In general, gas exchange rate (R,  moles m~2 hr"1) is a simple function
of the difference between partial pressure of the gas in the atmosphere
(Pg, atm) and the aqueous partial pressure (Pw):

          R = K .(Pg - Pw)                                                 (4)

K is an exchange rate coefficient (moles m~2 hr"1 atm"1).  In an open-water
system subject to wind stress, K is clearly a function of wind stress.  The
explanation for this relation between K and wind speed (V) is, according to
the model of Kanwisher (1963), that there is a boundary of a finite thickness
which varies in response to wind speed.

     We have measured 02 exchange in our microcosm tanks in the following
manner.  Flow through a tank with no contained organisms was stopped.  The
upwelling of water from the inlet was simulated by putting a submersible
bilge pump, outlet up, in the center of the tank and its pumping rate was
adjusted to approximate the ordinary rate of water flow.  The oxygen level of
water in the tank was elevated by bubbling with compressed gas; then that
bubbling was stopped and the loss of 02 from  the water was measured.

     The measured oxygen exchange rate  constant  for  the microcosms under a
variety of wind conditions averaged 43  mmoles m~2 hr"1 atm"1.  This rate is
near the no-wind value given by  Kanwisher  (1963) and Kinsey and Domm  (1974).
We  conclude  that the rim around  the margin  of the microcosm  tanks and  the
negligible fetch in  the  tanks  combine  to reduce  gas  exchange  to a minimum.
The observed exchange  rate constant is  sufficiently  small  that we can  ignore
this process in our  calculations.   It  is generally accepted  (e.g. Kanwisher,
1963)  that C02  gas exchange is small relative to 02, at  least in part  because
generally  less than 1 percent of the  C02  in  seawater is  present as free  C02.
The nominal  exchange rate coefficient  of 25 moles nT2  hr"1 atm"1 can be used
for C02;  the process can, again, be ignored in the microcosms.

G.   Relationship  of A02  to AC02

     Most  older studies  of coral-reef.metabolism relied  on measurements of  02
 changes,  assumed a metabolic  quotient  of,-1.0, and  calculated carbon  flux
 directly from 02 data  (either with  or  without a gas  exchange correction).
 Qasim  and Sankaranarayanan  (1970)  did  not  assume that  02/  C02 =  1.0,  but
 rather used  RytherTs (1956) suggested  "best value"  of  -1.2.   Smith  and Marsh
 (1973) examined 02 and C02 data for Enewetak  Atoll  and concluded that the
                                      16

-------
 best general relationship between 02 and C02 Is Indeed very near 1.  They
 noted that there was apparently a more scattered data set in the night
 samples, and they attributed this to problems in resolving C02 changes at
 night.

      We have measured 02 and COj in microcosm runs in which the dominant
 organisms were a fouling diatom community.  These data again yielded 02/
 values very near -1.0, with greater scatter in the nighttime.   For the most
 part, 02 is an easier measurement to automate than is COz; therefore we report
 02 here and assume that the appropriate metabolic quotient is  near -1.   This
 is almost certainly an oversimplification.  The quotient does  vary, and it is
 likely to be a useful environmental indicator in its own right.   We will not
 pursue that point further in this report.

      Carbon dioxide changes in seawater can be related to another important
 metabolic process associated with coral-reef systems:  calcification.   Al-
 though  we do not pursue C02 analyses in this report, considerable effort has
 been devoted during this investigation to the development of appropriate
 analytical procedures for measuring metabolic processes via the  C02 system.
 A publication derived from this study (Smith and Key, 1975), describes  the
 utility  of the C02 system as a metabolic record in calcifying  ecosystems,
 while another paper (Smith and Kinsey,  1978),- describes the  methodology
 for C02  measurements in seawater.   Under most conditions simulated in  the
 studies  reported here,  calcification has been very slow.   Subsequent investiga-
 tions (Jokiel,  1978)  have demonstrated'that coral  calcification  is  sensitive
 to  water mot-ion,  a variable  which ordinarily assures  low values  in  the microsms.

 H.    Replication and Reliability of  Metabolic Measurements

      In  essence,  the utility of any  environmental  variable as  a  legitimate
 and quantifiable measurement of environmental status  can be  reduced to  a
 single question:   Can the particular variable be easily used to  determine
 statistically  reliable  differences between environments  in order  to provide
 a practical means  of environmental assessment?  Some  environmentally valuable
 variables  are  sufficiently slow or difficult to measure  that they fail  the
 second half of  this  question; unfortunately,  many  promising measurements  of
 biological community  structure  fall  into  this  category.   Other easily
 measured variables  (including many of the  chemical parameters which have  legal
 status)  fail to  have  demonstrable environmental utility within statistical
 limits.  In general,  the  problem with using -such variables lies with the  lack
 of  experimental  strategy  associated with  their measurement rather than with
 lack of inherent utility  of  the variables.

     Where they  can be used, metabolically active chemical variables may be
 relatively easy  to measure.  In at least some environments, these variables
 also appear to have environmental significance.  In the following section, we
will examine patterns .of metabolic variation  within versus between sample
 treatments. The data presented  are aperies of 02 metabolic rates obtained
 for microcosm communities^
                                      17

-------
     We extend conclusions about the utility of oxygen metabolism in the
microcosms to Oa metabolism in many real world situations.  In those situa-
tions, gas exchange may override the metabolic effects of 02metabolism, so
CO* (including CaC03), inorganic nutrient, and other metabolic fluxes are
likely to prove more useful than 02.  Data from the lagoon of Canton Atoll
(Smith and Jokiel, 1976) provide a case of point.   In that study, we related
budgets of carbon, nitrogen, and phosphorus to physical and biological
gradients.  We suggest, based on the analyses presented in this report and
on the Smith/Jokiel studies, that careful mass-balance budgets of many
metabolically active materials are probably of more use as environmental
measurements than are the absolute values of these components.  Unless the
components in question are likely to be major environmental stimuli or
depressants as they deviate slightly from the norm, their rates of deviation
are likely to be more significant than their absolute deviation.
                                      18

-------
                                  SECTION V.

                        MICROCOSM COMMUNITY METABOLISM
      The purpose of the studies reported in this section is to examine the
 responses of microcosm community metabolism to changes in environmental
 conditions.   The experiments reported here use biological oxygen flux as the
 primary community metabolic measure and evaluate the metabolic changes that
 result from the following forcing functions:   light, substratum, nutrient
 levels, salinity,  and direct manipulations of community structure.   The
 results are by no means final and are not intended to establish legal or
 environmental guidelines.   Rather,  they are intended to develop and demon-
 strate analytical technique for microcosm studies.

      After examining  several approaches to the data analysis,  we have conclud-
 ed  that it is most useful  to present the results in terms of one-way or first-
 order effects.   Some  experiments were conducted in a matrix format  to look at
 synergistic  (or cross-product,  or second-order)  treatment effects.   Invariably
 the first-order effects dominated,  and higher order effects were barely (or
 not at all)  detectable above the error terms.   We have therefore chosen to
 present the  results as a series of  single-order experiments.   It is  obvious
 that more  detailed and exhaustive analyses of community metabolism responses
 to  environmental variability should be undertaken to define the higher  order
 effects which surely  exist,  and these analyses should also examine duration
 and frequency of perturbations  as well as  quality and intensity of perturba-
 tions.   Those examinations  proved to be beyond the  practical scope of  these
 studies> partly due to time  losses  from  funding interruptions.

A.    Light

      Shallow  benthos  communities  in tropical  environments  (as represented by
our microcosms) are ordinarily  dominated by plants,  so a forcing function
which  exerts major influence on the metabolism of these communities is light.
The absolute  light level has not been  controlled  (other than "off" or "on")
in  the  outdoor microcosms during  the course of the experiments reported here.
Neutral density filters have been used in other experiments to vary the rela-
tive  intensity between treatments.  Figure 3 is a typical curve of oxygen
metabolism and light level versus time of day.  Two of  the tanks (Tl and T3)
represent variously manipulated fouling  communities developed in tank micro-
cosms over a period of 38 days.  The third tank (T12) shows the calculated
flux associated with water flowing into a clean tank; this may be interpreted
as an evaluation of the "ambient water blank" plus the analytical error of
the method.
                                     19

-------
                 1200
 1800
I JUNE
                                                             1200
              Figure 3.  A.  Curve of ambient solar radiation over
                         24 hours.  B.  Die! response of oxygen flux
                         in two tanks with fouling communities (Tl,
                         ;T3) and one .tank with no significant bio tic
                         component (T12).

     Clearly, responses to light dominate metabolism in these microcosm
communities.  It is, therefore,  necessary first to calculate the metabolic,
response to light and then to examine either residuals within a particular
photic response pattern or differences between responses.

     Figure 4 presents four scatter diagrams of half-hour increments of solar
radiation versus oxygen metabolism.   The community in question was a
                                      20

-------
cc
2



  0.04

  0.02

  0.00

 •0.02
 0.04

 0.02

 0.00

-0.02



 0.04

 0.02

 0.00
    -O.O
*
            A
-0.008+0.0435-L
 ^
 cP
                        7304  Tl, rs 0.834
                        PRE-STRESS CONTROL
                     +
            B  &0=-0.004+0.0435-L
                  2            rt   Q
   7304 Tl, r =0.872
   STRESS I CONTROL
   	1	
                     -0.003 4 0.0594 -L
                        7304 Tl, r= 0.862
                        STRESS 2 CONTROL
                     4	1	
                   =-0.007 +0.0429 -L
                                       '°
                        '•
                       7304  Tl, r = 0.957
                       POST-STRESS CONTROL
                    0.5
           1.0
                                        1.5
                LANGUEYS PER MINUTE
 Figure 4.   Solar  ra'diation versus oxygen flux for
             four successive measurement periods in
             the same  tank.
                     21

-------
diatom-dominated fouling community which developed on rubble substratum and
tank walls over approximately one month previous to the first metabolism
measurements.  Water began flowing in the tanks on 9 June 1973, and diagrams
A through D represent data collected between 10 July and 31 July.  Several
observations can be offered.

     For this series, the linear relationship between light and oxygen metab-
olism is excellent (correlation coefficient varies from 0.83 to 0.96).
Nevertheless, there are suggestions that the functional relation between
light and oxygen metabolism is non-linear.  In all instances, the y-intercept
of the regression lines overestimates the calculated -0-light oxygen flux
(Table 2).  In diagrams A through C, the oxygen flux values at intermediate
light levels appear to be underestimated by the regression lines while the
high-light values tend to be overestimated.  Diagram C, which has only two
data points above a light value of 0.5 langleys per minute, appears to have
a slightly  (although not statistically significant) steeper regression slope
than the other diagrams, with approximately half of their light values
between 0.5 and 1.5 langleys per minute.  These observations are all consis-
tent with the generally accepted conceptual model that light becomes progres-
sively less limiting to community or plant metabolism at progressively higher
levels (e.g. Jassby and Platt, 1976).


       TABLE 2.  REGRESSION EQUATIONS (in the form A02 - a + b-Light)
                       FOR DATA PRESENTED IN FIGURE 4

                 Parameters ± 1 standard error unit.  For comparison
                 the directly observed 0-light oxygen fluxes ± 1
                 s.e. are also reported	
               s-e-                   a ± s.e.         0-light 02 flux ± s.e,
A
B
C
D
0.0435 ± 0.0051
0.0435 ± 0.0036
0.0594 ± 0.0084
0.0429 ± 0.0027
-0.00813 ± 0.00384
-0.00356 ± 0.00228
-0.00338 ± 0.00292
-0.00689 ± 0.00216
-0.01775 ± 0.00189
-0.01050 ± 0.00078
-0.00860 ± 0.00160
-0.01033 ± 0.00273
      We  therefore  subjected several data sets to  two other regression models.
 One  mo4el  used all data with non-zero light values and considered oxygen
 metabolism versus  the logarithm of light level.   The other model used all
 positive light and oxygen values and considered 1/02 versus I/light.  This
 latter transformation is analogous to the Burke-Lineweaver plot used in
 enzyme kinetics  and  can be used to calculate the  parameters of the  familiar
 Michaelis-Menton hyperbolic curve describing metabolic response to  a limiting
 substrate.
                                      22

-------
       With some  data sets,  one transformation or the  other did  improve  the
 statistical fit (as measured by  the correlation coefficient for the data  in
 their transformed format).  Ordinarily the  improvement was not dramatic;  in
 many  instances  the correlations decreased.   The linear fit was actually the
 best  in the plurality of cases examined.  Some data  sets showed a poor linear
 relation between 0^ and light.  Invariably  this characteristic proved to  be
 the result of random scatter which  was not  improved  by the transformations.
 We conclude that the routine descriptive analysis of metabolic response to
 light in the microcosms is best undertaken  by  linear regression.   We believe
 that  the functional relationship is non-linear but that examination of the
 data in detail  for the functional, relationships would  require more detailed
 case-by-case consideration than would  ordinarily be  justified  in routine
 assays  of metabolic responses to variables other than  light.    Resolution of
 this question is,  of course,  fundamental to understanding the energy fluxes
within  particular  ecosystems.

      According  to  an analysis of covariance  (Table 3)  the regression slopes
among the four replicate treatments of Figure  4 do not  differ  significantly
from one  another.   The regression heights do show significant differences.
Inspection of the mean oxygen fluxes adjusted  for varying mean light levels
demonstrates  that  treatment C is displaced substantially  above the  other
curves.    It has  previously been  pointed out that the functional relation
between light  and oxygen flux is probably non-linear.  Substantial  differences
                    TABLE 3.  ANALYSIS OF COVARIANCE FOR DATA IN TABLE 2 AND FIGURE A
   Lln«
             Treatment
                               i*2
Ixy
 Deviations from regression

 f     Zdyx2    M.S.
  A. Analysis of Cov»fiance

   1     control period 1    34    6.9322    0.3015     0.01883
   2     control period 2    48    8.2333    0.3583     0.020S3
   3     control period 3    19    1.0356    0.0615     0.00492
   4     control period 4    22    5.4767    0.2347     0.01098
   5         within
   6  regression coefficient
   7         common       123    21.6778    0.9560     0.05526
   8      adjusted Beans
   9         total        126    23.4467    0.9865     0.05559

  Significant difference saong regression coefficients?

  *3,U9 • M.S.6/M.S.5 - 0.79   Ho difference

  Significant difference among regression line heights?

  F3,122 " M.S.8/M.S.7 - 3.82   Significant st P • 0.012


  1.  Adjusted Me«n T's

                ?x - b (Ij - Xtot> -  0.0191 • 0.0441 (0.626 - 0.511)

                ?2 - * <*2 - *tot> "  °'0il* * 0-0441 (0.490 - 0.511)

                f3 - b (83 - Jtot) -  0.0127 - 0.0441 (0.271 - 0.511)

                ?4 - b (*4 - tto,.) -  0.0183 - 0.0441 (0.587 - 0.311)
                 0.0435
                 0.0435
                 O.OS94
                 0.0429
                 0.0441
 33
 47
 18
 21
119
  3
122
  3
125
                0.0140

                0.0187

                0.0233

                0.0149
0.005719
0.004936
0.001264
0.000923
0.012842
0.000256
0.013098
0.001226
0.014324
0.000108
0.000085
0.000107
0.000409
                                        23

-------
in the distribution of light values  sampled  could be expected  to be reflected
in the characteristics of the regression equations.  We  therefore suggest  that
this analysis of covariance indicates the regression slope  to be less  sensi-
tive to this non-linearity than the  curve heights.   Corroboration for  this
interpretation may be found in the comparison (and obvious  bias) between
directly calculated 0-light Oa fluxes and the Y-intercept values.   It  is
further evident that covariance comparisons among light  versus  02 equations
should first consider different light treatments as  a possible  explanation
for observed regression differences.  This possibility can  be unequivocally
dismissed only in the situation of between-treatment intercomparisons  under
identically varying light regimes^ some caution with the data should make
recognition of this possibility obvious.

     Figure 5 and Table 4 compare the pre-stress control treatment  of  Figure
4a  (Tl) with two similarly  treated tanks 
-------
                       TABLE 4. ANALYSIS OF COVAKIAN'CE FOR DATA IN FIGURE 5
Deviations from regression
Line Treatment
1 Tl
2 T7
3 110
4 vithln
5 regression coefficient
6 comnon
7 adjusted means
8 total
Significant difference among
F3,99 " M.S.5/M.S.4 - 0.09
t Ex2
34 6.9322
34 6.9322
34 6.9322


102 20. 7967

104 20. 7967
Ixy
0.3015
0.3002
0.2777


0.8793

0.8793
Iy2 b f
0.01883 0.0435 33
0.01949 0.0432 33
0.01709 0.0402 33
99
3
0.05541 0.0423 102
3
0.05559 105
Edy'x2
0.005719
0.006494
0.005970
0.018183
0.000052
0.018235
0.000180
0.018415
M.S.



0.000184
0.000017
0.000179
0.000060

regression coefficients?
Ho difference




     Figure 6 presents another  comparison of replicate treatments.  Two  tanks
(Tl, T2) were allowed to develop  fouling communities on the bare tank floor
and walls  over 38 days (24 April  through 1 June 1973).   Three other tanks
(T7, T8, T9)  similarly developed  over only 22 days.  Analyses of covariance
for these  data are presented in Table 5.  The duplicate 38-day treatments are
  cc
  x
  CO
  UJ
  _l
  o
  2
   •»
  X
   cvj
  O
-0.02
      -0.02
       0.04

       0.02

       o.oo
      •0.02
          0.0
                              7302 Tl
                                               0   22-DAY, 2
                        0« s -0.006+0.0492 • L
                              r =0.980
                                             -b-o
                                                     O    7302 T8
                                                       « -0.006* 0.0468 • L
                                                          r« 0.907
                         ,
                         2
                                        o cP
                       7302  T2
                     -0.006* 0.0498 • L
                       r » 0.903
..o.
   7302 T9
=-0.004* 0.0348-L
   r « 0.798
                       7302 T7
                     -0.005 + 0.0408 • L
                       r s 0,898
                                                 0.3          1.0

                                             LANG LEYS PER MINUTE
               0.5         1-0

           LANGLEYS PER MINUTE
                                            I.S
           Figure 6.  Solarradiation versus oxygen flux treatments discussed
                      in text.
                                      25

-------
                                     TABLE 5.  ANALYSIS OF COVARIA.VCE FOR DATA IN FICVXE  6
         Line
                  Treatment
                                                     Deviations  fron  regression
                                                     f      Edyx2        M.S.
A.
1
2
3
4
5
6
7
Coaparlnon of 33-day replicates ' '
n
T2
vithln
regression coefficient
cocoon
»d jus ted Beans
total
38
38


76

77
6.7933
6.7933


13.5867

13.5867
0.3351
0.3373


0.6725

0.6725
o. oi?:oo
0.02056


0.03776

0.03950
0.0492
0.0493


0.0495


37
37
74
1
75
1
76
0.000666
0.003809
0.004475
0.000000
0.004475
0.001740
0.006215


O.OOC060
0.000000
0.000060
0,001740

         Significant differences mtr.ong  regression coefficients?
           Fl,74 ' M.S.4/M.S.3 . 0.00    No difference
B.  Comparison of 22-day realleates

 1          "               38     6.7933
 2          T8               38     6.7933
 3          19               38     6.7933
 4        vithin
 5  regression coefficient
 6        common            114    20.3800
 7      -adjusted means
 8        total             116    20.3800
                                                       0.2764
                                                       0.3175
                                                       0.2372
                                                       0.8311

                                                       0.8311
                             0.01394
                             0.01803
                             0.01061
                             0.04258

                             0.04286
                                   0.0408
                                   0.0468
                                   0.0343
                                  0.0408
                                  37
                                  37
                                  37
                                 111
                                   2
                                 113
                                   2
                                 115
         Significant differences aaoeg regression coefficients?
           F2,lll " M.S.5/M.S.4 • 3.22    Significant at P - 0.04

         C-  Intereonpartson between pooled 38-day treatments and pooled 22-day treatments
          1        T1+T2
          2      T7+T8+I9
          3        vithln
          4  regression coefficient
          5        coonoa
          6      adjusted means
          7        total
 77
116
13.5867
20.3800
193    33.9667

194    33.9667
0.6725
0.8311
           1.5036

           1.5036
0.03950
0.04286
          0.08236

          0.08259
0.0495
0.0408
             0.0444
                                                                                76
                                                                               115
                                                                               191
                                                                                1
                                                                               192
                                                                                1
                                                                               193
         Significant differences among regression coefficient!?
           *1,191 " K.S.4/M.S.ft - 7.84     Significant at P -  0.006
                              0.002692
                              0.003191
                              0.002331
                              0.008214
                              0.000475
                              0.008689
                              0.000280
                              0.008969
0.006215
0.008969
0.015184
0.000619
0.015803
0.000230
0.016033
                              0.000074
                              0.000238
                              0.000077
                              0.000140
                              0.000079
                              0.000619
                              0.000082
                              0.000230
                                                                  ee,



£
another  even  better  than  Ly  track                           "1       "68  "^ traCk  °ne
                                                            26

-------
       0.06
CM
 Ul
 
-------
     Obviously the direct intertank comparisons eliminate metabolic varia-
bility which is not explained statistically by the light versus metabolism
regression.  Both the higher correlations and the ability to make comparisons
if (or when) light data are not available make such direct intertank compari-
sons very useful.  A disadvantage of these direct comparisons is that data
collected on different days (or otherwise not simultaneously collected)  cannot
be compared.  Moreover, the statistical basis for interpreting regressions on
data arrays with rigorously defined independent and dependent variables is
more final;, established than is the basis for intercomparison of two dependent
variables  (Ricker, 1973).  With correlations as high as those shown in Figure
7, this latter point is trivial.  However, not all such intercomparisons can
be expected to show this virtually perfect correlation.  When the data sets
permit, we generally prefer covariance analysis among data sets for which
regressions of metabolism versus light have been calculated as the analysis
of primary preference, with the direct coinparisons as supplementary analyses.

     Let us consider the energetic significance of these light versus metab-
olism analyses.  For the communities represented, the regression slopes are
near 0.045 moles 02 m~2 hr"1 langley"1 minute.  This slope term can be re-
duced to 7.5X10"5 moles Oa/kcal light.  If we assume that one mole of oxygen
production equals  (to a first approximation) one mole of organic carbon
production and that one mole of   organic carbon has a caloric value of about*
120 kilocalories,  then the light to fixed carbon energy conversion efficiency
is about 1 percent.  Because of light absorption through the water column and
because the light measurement includes light outside the photosynthetically
active spectrum, this figure underestimates the efficiency with which actually
available  light  is converted to organic carbon.  Odum  (1971) cites conversion
efficiencies  of  3  to 5 percent for the net production of intensively culti-
vated terrestrial  crops and only 0.5 percent as an average favorable condition.

     Both  the relatively high conversion  efficiency and the nearly linear
relationship  between available light and productivity  suggest  that these
fouling communities  (and, by extension, other  shallow  reef benthos communi-
ties) are  strongly light-limited.  Available data for  shallov water reef
communities  (summarized by  Smith, 1974) usually yield  gross production  to
respiration ratios near  1.0.  Unless the  respiration rates of  auch communities
drop concommitantly with production, reef communities  at moderate water depths
may be heterotrophic.  The  shift in production with water depth  should  follow
an exponential decay  curve  in parallel with  the vertical pattern of light
attenuation.   It might be postulated that a shift in community composition
with  depth could compensate for lower light  levels; the high conversion
efficiencies  suggest  that  there is little ecological margin for  such a  shift.

      It  is evident that  the association between extensive development of
 successful coral reef communities and clear waters is  more  than coincidental.
 Gross  community production  rates near 0.5 moles m"2 day'1  (Smith,  1974) with
 solar  input of not more  than 600 g cal cm~s  day"1  (Holmes,  1957) imply  aolar
 conversion efficiencies  similar  to the ones we have measured in the micro-
 cosms.   Waters which  are naturally or artificially turbid must significantly
 depress  organic carbon production of reef benthos.
                                      28

-------
 B.    Substratum

      The composition of reef communities is closely related to the nature of
 the substratum inhabited by these communities.   Organisms living both within
 the interstices of the substratum and on the substratum surface are influenced
 by  the size of the interstices and by the stability of the substratum.   Sub-
 stratum irregularity alters the surface area,  hence the "surface area index"
 (Dahl, 1973).   Substratum composition (e.g. grain size) relates to the ambient
 environment, and as such is an index of environmental conditions (e.g.  water
 motion,  water clarity) which may directly affect the composition of the biota.
 Coral reef environments can be found effectively spanning the range of
 possible substratum characteristics, so it is  important to quantify reef
 response characteristics to substratum composition.   Artificial (e.g.  dredging)
 or  natural (e.g.  storms) processes may dramatically alter substratum charac-
 teristics through either abrupt change or slow transition.

      This section considers the oxygen metabolism of shallow benthos communi-
 ties  in  direct response to the nature of the substratum.   The experiments
 explicitly eliminate secondary variables which  may be correlated with sub-
 stratum  type but which may affect metabolism independently of substratum.
 For example,  the substrata discussed here (mud,  sand,  rubble,  hard bottom)
 may be associated with particular water motion  regimes; and water motion may
 affect metabolism directly.   Mud,  especially in shallow water where it  is
 likely to be re-suspended,  is often associated  with  lowered water clarity or
 depressed salinity—either of which might alter metabolism.   The tanks
 eliminate these secondary effects.

      Figure 8  is  a series of  scatter diagrams of light versus  oxygen flux for
 three microcosm fouling communities.   These communities developed on rubble,
 sand,  and mud  bottoms  for 33  days;  community metabolism was  then measured on
 each  of  eight  days over the next  21 days.   Figure 8a presents  the same  data
 as  the diagram in Figure 4.   Several facts  emerge.   First,  linear regressions
 provide  relatively good fits  for  the data from  all three  communities.   Second,
 despite  the good  descriptive  fit,  the true  functional  relationship  is
 apparently non-linear.   This  interpretation can be made with most confidence
 for Figure 8a  (rubble).   Third, the 0-light intercept  is  overestimated  by the
 regression lines.  Analysis of covariance (Table 6)  demonstrates  that there
 are significant differences among the regression slopes.  - Further  analyses
 demonstrate that  the lines  for sand and mud (T2,  T3)^o not  differ  from one
 another.   The  pooled slope  for those substrata  is 0.0276, and  the  slope for
 rubble is  0.0420.

     Analysis  of  covariance allows  the comparison of these data with data
 collected  in another experimental series.   The data  represented  in  Figure 6a
 and 6b are  for  fouling  community development over 38 days  (about  the same
 time  span  as the Figure 8 daja) on  the walls,and  floor  of the microcosms.
 Because  the rubble substratuifl  has alretfdy been isolated as significantly
 different  from  sand and mud, and because  the hard bottom  tanks have apparently
higher metabolism/light coefficients  tnan the rubble tank, only  the rubble and
hard bottom are compared."
                                      29

-------
M
 10
 tu
 _J
 o
  CM
 O
 0.04

 0.02

 0,00

-0.02

-0.04

 0.04

 0.02

 0.00

-0.02

-0.04

 0.04

 0.02

 0.00

-0.02
                RUBBLE
      -0.04
                                 7304  Tl
                             =-0.004 + 0.0420-L
                                 r  -  0.862
    7304  T2
= -O.O01+0.029I
    r  = 0.827
                                               L  ,
                MUD
                                 7304  T3
                           02=-0.004 + 0.0261 • L
                                 r = 0. 810
          0.0
                 O.S
       1.0
1.5
                    LANGLEYS  PER  MINUTE
   Figure 8.   Solar radiacion versus oxygen flux of benthos
              communities developed on various substrata.
                            30

-------
                      TABLE 6. ANALYSIS OF COV'AKIAKCE FOR DATA IN nCURK 8
  Line
          Treatment
                              Ex2
_Deyiatlons_froni regression
 f      Idy-^2"    M.S."
1
2
3
4
S
6
7
8
rubble (Tl)
•and (12)
mud (T3)
within
regression coefficient
cooQion
adjusted Deans
total
126
126
126
378
360
23.4733
23.4733
23.4733
70.4200
70.4200
0.9870
0.6855
0.6118
2.2842
2.2842
0.05S83
0.02928
0.02430
0.10941
0.11328
0.0420
0.0291
0.0261
0.0324

125
125
125
375
2
377
2
379
0.014333
0.009261
0.008356
0.031950
0.003367
0.035317
0.003870
0.039187


0.000085
0.001684
0.000094
0.001935
  Significant difference among regression coefficients?

  P2.375 ' M.S.j/M.S.4 - 19.81    Significant at 0.001
      Table 7 demonstrates that the  slope of the hard-bottom metabolism curve
 is steeper than the rubble metabolism curve.  Because  these experiments were
 not conducted  simultaneously, this  comparison must be  treated with some
 caution.  Nevertheless, the two experiments suggest a  metabolic progression
 with respect to substratum:  hard > rubble > sand « mud.   The data at hand
 suggest that the ratios of metabolic  response are 1.8:1.5:1.1.  Rubble and
 hard bottom are capable of supporting 50 to 80 percent higher metabolic rates
 than sand or mud as a direct consequence of substratum type.
—

Line Treataent
1
J
•(
4
1
6
7
rubble
hard
within
regression coefficient
adjusted Beans
Co tat

f
126
77

203
204

Ex2
23.4733
13.3867

37.0600
37.4278

i*jr
0.9870
0.6725

1.6594
1.6714

V
0.05583
0.03950

0.09533
0.09571
Deviations from regression
b
0.0420
0.049$

0.0448

f
125
76
201
• -1
202
1
203
Idy'x2
0.014333
0.006215
0.020548
0.'000477
0.021025
0.000043
0.021068
H.S.


0.000102
0.000477
0.000104
0.000043
  Stgoifleant difference between regression coefficients?

  *1.301 " M.S.4/K.S.3 - 4.68    Significant «t 0.03
      The Intercomparison of metabolic rates  can be effected more directly
for  the mud, sand, and rubble substrata.  Figure 9 presents scatter diagrams
of rubble versus sand and rubble versus mud  plots and the functional or
geometric mean regression coefficients (after Ricker, 1973).   If those slopes
are  used to establish the rubble:sand:mud metabolic ratios, the  results are
1.5:1.1:1—virtually identical with the covariance approach.

     The explanation for enhanced metabolism on rubble or hard-bottom sub-
strate  is not entirely clear.  Experimental  design rules out water motion or
                                        31

-------
'oc     °'04

M

*2

 to —  0.02
 UJ O
 0<
 CM
O
<3

CM
H
       0.00
      -0.02
       0.04
 V
  to-  0.02
  Ul O
  02
   M
  O
  10
        0.00
       -0.02
       7304

T2 = 0.001 + 0.724 Tl

r = 0.983
               1	1	1	1	1	1	1	h

                       o       7304
                       T3 = -0.002 + 0.6 60

                        r  =0.927
                '     '    •*••
                                      *    * - ••- *
              -0.02    0,00     0.02     0.04     0.06
                         ,  MOLES  M~2HR"~'
                          (RUBBLEV
    Figure 9.  Direct comparison of oxygen flux on rubble
              versus sand and mud suoscrata.
                        32

-------
 water  clarity.   Physical stability may be a partial answer, inasmuch as bio-
 turbation of  sand and mud could  disrupt  the algal communities on those sub-
 strata.   Increased surface area  for algal growth could explain the difference
 between  the rubble and the sand-mud comparison; this explanation is not
 consistent with  the low-surface  area hard bottom being the most active sub-
 stratum.   There  may be an effect associated with reflectivity of the surfaces,
 since  light does appear to be a  limiting resource.  Bacterial respiration in
 the sediments is yet another possible  explanation.  Without having an
 entirely  satisfactory explanation for  this phenomenon; we report the existence
 of a direct link between type of substratum and metabolic activity in these
 shallow tropical benthos microcosms.

 C.   Nutrients

      Coral reefs are ordinarily  considered to be products of low-nutrient
 conditions, and  to cycle available nutrients effectively.  Kaneohe Bay,  Hawaii
 (the site of our microcosm facility),  is subjected to high levels of nutrient
 loading,  primarily from sewage discharge.  We therefore wished to determine
 whether reef communities in Kaneohe Bay were still limited by available
 nutrients or were nutrient-saturated.

      Fouling communities were allowed  to develop in each of three tanks.   The
 experimental communities also included three grazing herbivorous  fishes
 (.Aoanthucus trioetegus) in each tank.   After the fouling communities  were
 inoculated with unfiltered water from  the bay,  bay water flow  to  one  tank was
 cut by 30 percent (i.e. by 3 liters  min'1).   Total flow was brought back to
 10 liters min"1  by the addition of water  from a saltwater well.   Salinity  of
 that well water  is virtually identical to salinity of bay water  (" 35 °/oo),
 but nutrient  levels of the well water  are substantially  higher than bay water
 (Table  8). Thus, there were two  replicate bay  water communities  compared  with
 one community  enriched with well  water.
TABLE  8    COMPARISON OF TYPICAL DISSOLVED  INORGANIC NUTRIENT LEVELS IN BAY
             WATER DRAWN THROUGH MICROCOSM  FACILITY WITH SEAWATER WELL
                          PC*
           NO:
             NHL
        Total inorganic N
                                           ymoles/liter
Bay water
Well water
30% well + 70% bay
Well/bay ratio
0.5
1.2
0.7
2.4
0.6
5.0
1.9
8.3
1.6
5.0
2.6
3.1
2.2
10.0
4.5
4.5
(70% bay +302 well)/
bay ratio
1.4
3.2
1.6
2.1
                                      33

-------
     Figure 10 illustrates metabolic rate  as a function of light in the three
tanks.   The analysis of covariance is presented in Table 9.  It is evident
that the 30 percent bay water addition boosted community metabolic response
to light by about 43 percent and that the  two replicate treatments receiving




X
CM
1
CO
111
-J
O
2.
X*
1
_J
U.
o





0.06

O.O4
0.02
0.00
-0.02
0.06
0.04
0.02
0.00
-0.02
0.06
0.04
0.02
0.00
-0.02
0
	 ,— 	 	 1 	 ; 	 '
A r = 0.846

-------
                      TABLE 9.  ANALYSIS OF COVARI/NCE FOR DATA IN FIGURE 10
                                                            Deviations from regression
  Line
          Treatment
                              Ex*
      Zxy
              Idyx*
                                            M.S.
1
2
3
4
5
6
7
8
bay water 1
bay + well
bay water 2
within
regression coefficient
common
adjusted means
total
68
68
68

204

206
7.8533
7.8533
7.8533

23. 5600

23.5600
0.2638
0.3670
0.2490

0.8798

0.8798
0.01237
0.02359
0.01234

0.04830

0.04971
0.0336
0.0467
0.0317

0.0374


67
67
67
201
2
203
2
205
0.003509
0.006439
0.004447
0.014395
0.001051
0.015446
0.001410
0.016856



0.000072
0.000525
0.000076
0.000705

  Significant difference among regre«»lon coefficients?

  F2,201 ' M.S.5/M.S.4 " 7.29   Significant at 0.001
bay water tracked one another well.   Figure 11 is a  plot of 02 metabolism  in
the bay + well tank versus 02 in the two bay tanks.   This direct fit yields  i
well/bay  regression slope of  1.36—very close to the value of 1.43 inferred
from  the  covariance analysis.  The metabolic enhancement by well water is
close to  the proportional increase in phosphorus loading (Table 8).
                        0.04
               X
              CM
               
               3
               o
              O
              <3
                        0.02
                        0.00
                      -0.02
   T5  •
   Til  o
             7305
            T7= 0.003+ 1.35715,1
             r =0.980
-0.02
                                          0.00
0.02
0.04
                              T5.ll  A02,  MOLES  M"2HR"'
                                             (BAY)
              Figure 11.  Direct  comparison of "low nutrient" (bay
                           water)  versus "high nutrient" (bay4well
                           water)  treatments.
                                      35

-------
     There are several conclusions to be drawn from these analyses.  The
microcosm benthos—and by Inference the shallow benthos of Kaneohe Bay—
increase their metabolic rate when nutrient levels are raised.  Therefore,
increased nutrient loading in the bay would result in higher benthos produc-
tion.  Conversely, lowering the nutrient loading should decrease production.
Moreover, the microcosms prove from these preliminary observations to be a
useful tool for examination of the effects of nutrient loading.  At this
writing, this set of conclusions has been coupled with the fact that a major
nutrient input to Kaneohe Bay is being removed (diversion of sewage), to
expand this observation into a more detailed analysis of the effects of
nutrient loading on reef communities.  Included in that analysis is an assess-
ment both of the generalized effects and a more specific analysis of the
individual roles of nitrogen and phosphorus.

D.   Salinity

     Reef ecosystems are ordinarily restricted to areas of near-oceanic salin-
ity  (* 35 °/oo), although there are reefs which experience moderate departures
from this value.  Freshwater "kills" from rains and runoff have been reported.
Reef communities are frequently suppressed in the vicinity of streams.  The
tolerance of reef communities for salinities between 0 °/oo and. 35 °/oo has
not been well defined.  We therefore conducted an experiment with two primary
purposes.  Approximately what salinity depression begins to kill conspicuous
parts of the reef community?  Does a measurable metabolic response occur at
salinities intermediate between 35 °/oo and this lethal threshold?

     Fouling communities were allowed to develop on rubble, sand, and mud
substrata in the microcosms.  The results within the substratum types were
qualitatively similar, but between-salinity differences were most conspicuous
on the rubble substratum.  Only the results from the rubble experiments are
reported here.  Organisms listed in Table 10 were added to the fouling


          TABLE 10.  ORGANISMS INTRODUCED INTO FOULING COMMUNITIES
	FOR SALINITY STRESS EXPERIMENTS	

ALGAE                                    CRUSTACEANS
  Aoanthapora spieifera                    Chiridota Twwaiiena-Ls

ECHINODERMS                              FISHES
  Opheodesoma speotabatis                  Acanthums triost&gus
  Holothiwia monocaria

CQELENTERATES
  Zoanthus paeificus
  Poeillopova damicornis
  Pori-tes eompresea
                                      36

-------
 communities.  These organisms are all typical of the fringing reef communities
 of Kaneohe Bay.  It was determined that the organisms added were themselves
 insufficient to alter the microcosm community metabolism significantly.
 Hence, observed metabolic responses—if any—would be that of the fouling
 communities.  These large organisms were simply added as indicators of stress.

      Table 11 reports the chronology of salinity alteration.  The initial
 salinity reductions (to 29 and 25 °/oo) imposed no measurable metabolic
 response or lethal effect on the microcosms, so the stresses were relaxed and
 then applied more severely (to 22 and 16 °/oo).   The experiment was performed
 as described, with the assumption that the initial salinity reductions had no
 effect.  This experiment should be repeated in a manner which precludes the
 need for such an assumption.


 	TABLE 11.   CHRONOLOGY OF SALINITY STRESS EXPERIMENTS

 Date (day of year)             Day of                 Event
 	Experiment	

 11 June 1973  (162)              0          Start experiment

 14 July 1973  (195)             33          Retain tanks 1,2,3 as controls
                                             Lower tanks  7,8,9 to 25 °/oo
                                             Lower tanks  10,11,12 to 29 °/oo

 19 July 1973  (200)             38          Return all tanks to 35 °/oo

 24 July 1973  (205)             43          Retain tanks 1,2,3 as controls
                                             Lower tanks  7,8,9 to 22 °/oo
                                             Lower tanks  10,11,12 to 16 °/oo

 25 July 1973  (206)             44          Return all tanks to 35 °/oo

 17 August  1973  (229)             67          Terminate experiment
     Only  the lowest salinity  (16 °/oo) killed a substantial fraction of the
reef organisms.  All organisms listed in Table 10 as well as some of the
infauna which had entered the  microcosms as fouling organisms were affected
during the 24-hour exposure.   The coelenterates and echinoderms were all
killed by  the lowest salinity;  the  other animals recovered once the salin-
ity  was  returned to normal.  Algal survival could not be assessed by visual
inspection, so is not reported here.  Figure 12 and Table 12 show the results
of the metabolic measurements.  Over the entire salinity range employed, there
was no significant variation in metabolic response.
                                      37

-------
 M


  CO
  UJ

  O
  2
   m
  X

  _l
  U.
   CM
  O
 0.04

 0.02

 0.00

-0.02



 0.04

 0.02

 0.00

-0.02



 0.04

 O.02

 0.00
                      0     7304  Tl
                     A02 = -0.004 + 0.042O • L
                            r = 0.862
29%
           73O4  TIO
        = -O.OO5* 0.0387 L
           r = 0.850
      -0.02
                             7304 T7
                             (.003 + 0.0
                             r = O.87I
                A02 = -0.003 + 0.0429- L
                                             7304  T7
                                     A02 = -0.004 + 0.0561 - L
                                             r =  0.812
                                     I	1	
   7304 TIO
= -0.004+0.0399-L
   r = 0.758
                                    O.S          IJO
                                 LANGLEYS PER  MINUTE
                 1.5
         0.0         0.5          1.0          I.!
                 LANGLEYS PER MINUTE
          Figure 12.  Solar radiation versus oxygen flux under various
                      salinity regimes.
      There are two primary conclusions to be drawn from  this  surprising re-
sult.   Although salinities below 22 °/oo (i.e. below 63  percent of ambient
salinity)  did prove detrimental to the conspicuous reef  organisms that were
added to  the microcosms,  these organisms were minor in the  total metabolic
activity  of the reef community.  If we assume that the effects  of lowered
salinity  were ecologically significant—despite their relatively minor
"energetic" importance—then we conclude that there was  not a community
metabolic harbinger to damage from salinity depression.  Although it has been
demonstrated that community metabolism is'responsive .to other  perturbations,
metabolism of reef benthos communities is not sensitive  to  sublethal varia-
tions in  salinity.
                                      38

-------
                       TABLE 12. AKAI.YSIS OF COVA&IAXCE FOR MTA Iff FIGURE 12
    Line
           Treatment
                               1*2
                                      Zxy
                                                             Deviations from regression
                                                             f     Edyx2      M.S.
1
2
3
4
s
6
7
a
9
10
35 °/oo
29 °/oo
25 °/oo
22 °/oo
16 °/oo
within
t«gre*6ion coefficient
comnon
adjusted means
total
126
48
48
19
19

260
264
23.4711
8.2333
6.2333
1.0356
1.0356

42.0089
43.8400
0.9669
0.3175
0.3521
0.0582
0.0413

1.7561
1.8197'
0.05583
0.01696
0.01984
0.00497
0.00287

0.10047
0.10316
0.0420
0.0387
0.0429
0.0561
0.0399

0.041*

125
47
47
18
18
255
4
259
4
263
0.014332
0.004715
0.004781
0.0001695
0.001223
0.026746
0.000314
0.027060
0.000566
0.027626





0.00010
0.00007!
0.00010'
o.ooou:
   Significant differences among regieisiou coefficients?

   F«,255 - M.S.7/M.S.6 - 0.7S   No difference
 E.    Biological Manipulation

      The analyses presented  so  far have taken little note of  the  biological
 composition of the microcosm communities.  The communities have been considered
 only  as   fouling communities, with reference to some specific community
 alterations common to all  treatments.   Such a presentation has been deliberate.
 The  philosophy of the analyses  and of the presentation has been that com-
 munity metabolism is a) easy to measure,  b) sensitive to variation  in (some)
 external forcing functions,  and c) relatively insensitive to  uncontrolled
 variations in community structure  associated with random recruitment.

      The previous sections have established the validity of these points.  Is
 community metabolism sensitive  to  direct, deliberate manipulation of biologi-
 cal  composition?  This question is the subject of the present section.

      Three tanks were set up with  a sand  substratum and allowed to  develop a
significantly different feeding strategies,



respiration of the fishes themselves was a negligible contribution to the
total microcosm community metabolism.

     Figure 13 is a plot of oxygen metabolism versus
treatments, and Table 13 is the covariance analysis.
                                                     a
is
              to  fish
                                        given compar ab !*
latlon affects metabolism.
                               ^^
                                       39

-------
I
 cc
 I
CM
I
 (O
 UJ
U-

 CM
O
      0.06
     -0.02
       0.06
      0.04
      -0.02
      0.06

      0.04

      0.02

      0,00

     -0.02
                                      
-------
                         TARLF 11  AVH.YSIS Or COVASIA^CE FOR DATA IS FIGURE 13
            Treetatnt
                                 1*2
                                                      Depletion* frota Regression
                                                      "7"    Idyx*M.S.
1



0 fifth, **nd
3 Aaantrsu'ua, end
3 Soarus, «»nd
within
zecretdon coefficient
COSQOB
adjusted Bean*
total
68
68
68
204
206
T.8S33
7.8533
7.8333
23.5600
23.5600
0.3314
0.2230
0.1757
0.7301
0.7301
0.01858
0.00908
0.00610
0.03376
0.3519
0.04230
0.0285
0.0225
0.0310
67
67
67
-201
2
203
2
205
0.004593
0.002750
0.002169
0.009512
0.00162!
0.011135
0.001430
0.01256S


0.000047
0.000811
0.000055
0.000715
                                       0.3987

                                       0.3987
Bepeet llaec 4-8 for Aaant'airus v. Saearut

 4'       within
 5'  regret*ion coefficient
 «'       coran       136   15.7067
 7'    «djv«t«d mean*
 «'       t«Ml        137   15.7067

Signlflciut difference* u»ng regression coefficient*?

'2,201 • K.S.j/H.8.4 - 17.26   Significant *t P « 0.001

Bapeit. tilting 4oenthun«« v. Soarut

                                  t P - 0.05
                                     0.01518

                                     0.01527
                                                       0..025S
134
 1
135
 1
136
0.004919
0.000142
0.005061
0.000090
0.005151
0.000037
0.000142
0.000037
0.000090
      The  next  comparison involves a complex reef  rubble  community (trans-
planted from the reef  flat nearly in toto  and then allowed to  stabilize), a
sterile rubble community to which two species of  grazing echinoids (5  speci-
mens each of adult Tripneustes gratilla and Echvnom&tiea  matfaei)  were  added,
and the duplicate rrt>bl&/Acanthia>u8 communities previously considered.
Analysis  of covariance (Table 15) demonstrates no significant  difference
among the regression slopes.   The variety  of choices of  grazers,  from  fishes
to  large  invertebrates to cryptic invertebrates,  has imposed similar effects
on  community metabolic response to  light.   Apparently community metabolism
is  sensitive to grazing pressure but not to the quality  of grazing.  Either
by  chance or adaptive  response, the communities have shown very al>U«
responses despite our  approach to adding the herbivore level in a trophic
pyramid.
                     TABLE ™.  AMtLTSIS OF OOVMtlAKCE. KO FISH VERSUS FISH. KTOBLj
                                                                  Deviations from regression
  Line
Treatment
   1      0 fish, rubble     137
   2   3 Aoar.thurua, rubble   137
   3        within
   4  tegrenion coefficient
   J        coinoa         274.
   6      *dju«ted «ean«
   7        total         275
15.7067
15.7067

31.4133
» 
-------
     There does  appear to be a substantially lower net production in the reef
rubble community than in the two simpler communities  (Table 15b).  The three
communities respond similarly to varying light,  so we assume that the lower
net production represents a higher community respiration rate.


                TABLE 15.  ANALYSIS OF COVARIAKCE. REEF RUBBLE V. ECHIKOIPS V. FISH               	.
                                                          Deviations from regression
Line Treatment
A.
1
2
3
4
5
6
7
8
Analysis of Covatiance
reef rubble
sterile rubble 4- echinoid
sterile rubble + fish
within
regression coefficient
coumon
adjusted means
total
f

68
68
137


273

275
£X2

7.8533
7.8533
15.7056


31.4122

31.4122
Ixy

0.3054
0.2542
0.5128


1.0724

1.0724
ryz

0.01599
0.01243
0.02477


0.05319

0.05389
b

0.4390
0.0324
0.0327


0.0342


f

67
67
136
270
2
272
2
274
Zdyx*

0.004111
0.004200
0.008029
0.016340
0.000236
0.016576
0.000700
0.017276
M.S.




0.000061
0.000118
0.000061
0.000350

 Significant differences among regression coefficients?

 *2,270 " M.S.5/M.S-4 - 1.94   Ho difference at P - 0.10


 B.  Adjusted aean T*« 1 95Z confidence interval

          Yj - 0.0024 1 0.0019

          ?2 - 0.0069 * 0.0019

          23 - 0.0052 ± 0.0013
      One might anticipate, a priori,  that the effects of grazing pressure  on
reef  metabolism would have been different from those that we observed.   In
some  situations, cropping seems to accelerate growth; that proves not  to be
true  for reef community metabolism, although it might be the case for  the
metabolic rate normalized per unit biomass rather  than per unit area.   This
latter normalization is difficult  (in fact,  apparently impossible without  at
least partial destruction of the community).

      We conclude that community metabolism is indeed sensitive to biological
manipulations within the microcosms.   We suggest that there may be relatively
broad limits of herbivore food uptake requirements over which the community
metabolism adjusts to a constant rate.  A sufficiently low herbivore  food
demand apparently does allow the community metabolism to Increase, and an
excessive herbivore food demand will  result in eventual starvation.  Although
not explicitly examined by these analyses, we would anticipate analogous
patterns among carnivores and detritivores.

F.    Conclusions

      In the preceding sections we have examined the metabolic responses of
shallow tropical benthos communities  to five forcing functions:  light, sub-
stratum, nutrient loading, salinity,  and direct manipulation of the biological
community.  The first three forcing functions may be treated conceptually  as
limiting resources, with light being  a primary limitation which has not been
controlled in these experiments.   Salinity depression was Investigated as  a

                                       42

-------
direct stress imposed upon the community.  Manipulation of community structure
and, inferentially, metabolic similarities in apparently replicate treatments
provide insight into the ability to extrapolate from the metabolic character-
istics of such simplified simulations of nature to reefs under natural
conditions.

     The manipulations and replications demonstrate that the communities are
sensitive to biological alterations but that the sensitivity is only between
distinctly different communities.  Treated similarly, the communities converge
towards similar metabolic rates.  Moreover, complexity of the manipulated
communities does not appear to be a major factor in the metabolic response
characteristics.  On this basis, we may extrapolate from the microcosms to the
real world.

     Limiting resources in the microcosms are strongly reflected in their
metabolic responses.  Light, substratum characteristics, and nutrient loading
all impose substantial and quantifiable metabolic responses on the microcosm
communities.  We conclude that any alteration of these variables will impose
effects on reef communities.  It is likely that, as low-level chronic pertur-
bations, these variables are altered more frequently in tropical nearshore
ecosystems than other variables.  The data presented here only begin to
reduce our ignorance about metabolic responses to these variables; far more
work is needed to quantify these responses adequately.  Information on light
response is of primary importance and accumulates most rapidly in an experi-
mental facility such as we describe, because all experiments so-conducted
will contain light as an "uncontrolled" independent variable.  By such
treatment in outdoor microcosms, we are able to achieve natural levels of
intensity and natural photoperiod.  We are continuing to define reef benthos
community responses to nutrient loading.  Metabolic responses to substratum
alteration remain inadequately described.

     The remaining variable which we have examined is salinity.  This vari-
able is also frequently altered in nearshore environments.  Reef communities
have an unexpected resilience to salinity depressions.  Neither structural
nor metabolic responses to depressed salinity occurred in the microcosms
untifsalinity was reduced below 22 %>o.  Thus, the deleterious effects to
Telfs associated with lowered salinity may be attributed to extreme salinity
depression (probably largely in the form of freshwater lenses floating on the
seawaSr) or to materials introduced with the fresh water (sediment, nutri-
enta? toxins).  Salinity depression  does not appear to be a m*jor chronic
sublethal stress on reef communities.
                                     43

-------
                                 REFERENCES

Anonymous.  1976.  Standard methods for the examination of water and waste-
     water, American Public Health Association, Washington, D. C.  1193 pp.

Beyers, R. J.  1963.  A characteristic diurnal metabolic pattern in balanced
     microcosms.  Publ. Inst. Mar. Sci.  9:19-27.

Confer, J. L.  1972.  Interrelations among plankton, attached algae, and the
     phosphorus cycle in artificial open systems.  Ecol. Monogr.  42:1-23.

Cooke, G. D.  1967.  The pattern of autotrophic succession in laboratory
     microcosms.  BioScience, 17:717-721.

Dahl, A. L.  1973.  Surface area in ecological analysis:  quantification of
     benthic coral reef algae.  Mar. Biol., 23:239-249.

Hall, D. J., W. E. Cooper, and E. E. Werner.  1971.  An experimental approach
     to the dynamics and'structure of freshwater animal communities.  Limnol.
     Oceanogr.  15:839-928.

Henderson, R. S., S. V. Smith, and E. C. Evans III.  1976.  Flow-through
     microcosms for simulation of marine ecosystems:  development and inter-
     comparison of open coast and bay facilities.  Technical Report NUC
     TP519.  Naval Undersea Center, San Diego, California.  80 pp.

Holmes, R. W.  1957.  Solar radiation, submarine daylight, and photosynthesis,
     In:  Treatise on Marine Ecology and Paleoecology Vol. 1.  ed. by
     J. W. Hedgpeth.  Geol. Soc. of America Memoir 67, pp. 109-129,
     Washington, D. C.

Jassby, A. D., and T. Platt.  1976.  Mathematical formulation of the
     relationship between photosynthesis and light for plankton.  Limnol.
     Oceanogr.  21:540-547.

Jokiel, P. L.  1978.  Effects of water motion on reef corals.  Jour. Exp.
     Mar. Biol. Ecol., 35:87-97.

Jokiel, P. L,, S. L. Coles, E. B. Guinther, G. S. Key, S. V. Smith, and
     S. J. Townsley.  (manuscript)  Effects of thermal loading on Hawaiian
     reef corals.  Final report of EPA .grant 18050DDN.  Submitted to
     D. K. Phelps, National Marine Water Quality Laboratory, Narragansett,
     Rhode Island, November 1974, 285 pp.
                                     44

-------
Kanwisher, J.  1963.  On  the exchange of gases between  the atmosphere and the
     sea.  Deep-sea Res., 10:195-207.

Kinsey, D. W.  1973.  Small-scale experiments to determine the effects of
     crude oil films on gas exchange over the coral back-reef at Heron Island.
     Environ. Pollut., 4:167-182.

Kinsey, D. W., and A. Domm.  1974.  Effects of fertilization on a coral reef
     environment - primary production studies.  Proc. 2nd Internat. Coral
     Reef Symp.  1:49-66.

Mullin, M. M., and P. M.  Evans.  1974.  The use of a deep tank in plankton
     ecology.  II:  Efficiency of a planktonic food chain.  Limnol.
     Oceanogr.  19:902-911.

Nixon, S.  1969.  A synthetic microcosm.  Limnol. Oceanogr.  14:142-145.

Northby, J. A.  1976.  A  comment on rate measurements in open systems.
     Limnol. Oceanogr.  21:180-182.

Odum, E. P.  1962.  Relationships between structure and function in the
     ecosystem.  Jap. Jour, of Ecol., 12:108-118.

Odum, E. P.  1971.  Fundamentals of ecology, 3rd ed., W. B. Saunders Company,
     Philadelphia, Pa.  574 pp.

Qasim, S. Z., and V. N. Sankaranarayanan.  1970.  Production of particulate
     matter by the reef on Kavaratti Atoll (Laccadives).  Limnol. Oceanogr.
     15:574-578.

Ricker, W. E.  1973.  Linear regressions in fishery research.  Jour. Fish.
     Res. Bd. of Canada,  30:409-434.

Ryther, J. H.  1956.  The measurement of primary production.  Limnol.
     Oceanogr.  l(2):72-84.

Ryther, J. H., W. M. Dunstan,  K. P.  Tenore,  and J. E. Hugunin.  1972.
     Controlled eutrophication increasing food production from the sea by
     recycling human wastes.   BioScience, 22:144-151.

Smith,  S. V.   1974.  Coral reef carbon dioxide flux.  Proc. 2nd Internat.
     Coral Reef Symp.  1:77-85.

Smith,  S. V., and G. S.  Key.   1975.   Carbon dioxide and metabolism in marine
     environments.   Limnol. Oceanogr.  20:493-495.

Smith,  S. V., and D. W.  Kinsey.  1978. , .Calcification and organic carbon
     metabolism as indicated by carbon ^dioxide.   In:  Coral Reefs:  Research
     Methods, ed. by D.  R>Stoddart arfd R.  E.  Johannes.  UNESCO,  Paris.
     581 pp.
                                     45

-------
Smith, S. V., and F. Pesret.  1974.  Processes of carbon dioxide flux in the
     Fanning Island lagoon.  Pac. Sci., 28:225-245.

Smith, S. V., and J. A. Marsh, Jr.  1973.  Organic carbon production on the
     windward reef flat of Eniwetok Atoll.  Limnol. Oceanogr.  18:953-961.

Smith, S. V., and P. L. Jokiel.  1976.  Water composition and biogeochemical
     gradients in the Canton Atoll lagoon.  In: An Environmental Survey of
     Canton Atoll Lagoon 1973, ed. by S. V. Smith and R. S. Henderson, Naval
     Undersea Center Technical Report NUC TP 395, San Diego, Calif., pp. 5-14.
     Also as Atoll Research Bull. No. 221, Smithsonian Inst., Washington, B.C.

Taub, F.  B.  1969.  A biological model of a freshwater community:  a
     gnotobiotic ecosystem.  Limnol. Oceanogr.   14:136-142.
                                    46

-------
                                  APPENDIX

                      DESIGN,  CONSTRUCTION, AND OPERATION
               OF  SHALLOW TROPICAL BENTHOS MICROCOSM FACILITIES
A.   General
     Two microcosm facilities were developed during the course of this pro-
gram.  These facilities are outdoor constructions designed to simulate
environmental  characteristics typical of shallow tropical benthic settings.
The prototype  (Figure 1) is at the Hawaii Institute of Marine Biology (HIMB)
and draws water  from protected, organic-rich Kaneohe Bay, Hawaii.  This
facility evolved gradually.  All experiments described in this report were
performed at the HIMB facility.  The Naval Undersea Center (NUC, now the
Naval Ocean Systems Center) used the HIMB experiences to design and establish
an analogous facility on the clean-water open coast of nearby Ulupau Head.
Experiments began there near the termination of the contract period covered
by this report.

     Despite the proximity of the two locations to one another (Figure A-l)
and the similarity of "operating philosophy" at the two facilities, site-
specific physical, chemical, and biological considerations provide useful
insight into the design and operation of such microcosm facilities.
Henderson et al. (1976) describe the NUC facility in some detail and provide
results from preliminary inter-calibration experiments.. This appendix
presents design, construction, and operating considerations consistent with
the two localities.  Inasmuch as design detail must inevitably be site-
specific and "state-of-the-art" specific, plans and layout are not presented
in any detail here.

     The primary operating constraint at HIMB is associated with biological
fouling.  Large  numbers of marine organisms settle throughout the seawater
system and capitalize on abundant particulate organic food to grow rapidly,
clog pipes,.and  restrict water flow.  The system was designed and constructed
to minimize this problem in a manner which in no way impinges upon biological
activity within  the microcosms themselves.  At NUC, the primary operational
hurdle was associated with inexpensively securing a seawater system which
could reliably draw ocean water from beyond a surf zone where breaking winter
waves commonly exceed 3 meters in height for days to weeks at a time.

     We describe the microcosm components according to the following
divisions:
                                      47

-------
                       Kaneohe  Bay
B.
Figure A-l.   Map showing locations of HIHB and NUC microcosm facilities.


 a.  seawater supply  and distribution system;

 b.  microcosm aquaria and associated devices for environmental
     modification;

 c.  automatic sampling and data acquisition system.

 Seawater Supply and  Distribution System
     The microcosms must be supplied with an uninterrupted flow of seawater.
The water should be free of uncontrolled contaminants (e.g. fresh water or
sediment from storm runoff). Delivery of water should exceed total antici-
pated flow through the microcosms by at least 50 percent, so that there is
an adequate supply for overflow  from constant head tanks, for holding tanks,
and so forth.  An adequate system should meet the following criteria:

     1.  Inertness of material—-All surfaces coming in contact with water to
be delivered to the system must  be made of non-toxic plastic, fiberglass, or
titanium.  All wetted surfaces should be conditioned with running seawater
                                    48

-------
 for several weeks  to  leach out  any  residual  solvents or other possible toxins
 before  experimental organisms are added.  Folyvinyl chloride (PVC) pipes,
 valves,  and other  fittings are  acceptable, and PVC is particularly easy to
 use and modify.  A single  toxic (e.g.  copper) fitting among many square
 meters  of  otherwise inert  materials may prevent  the survival of delicate
 organisms  (e.g.  corals).

      2.  Reliability—Microcosm experiments  may  run for months and are
 vulnerable to  interruptions in  seawater delivery.  The entire delivery
 system  should  be constructed in duplicate (see below, on cleaning) so that
 there is adequate  system redundancy in the event of system failure.  The
 primary  pump for the  facility is likely to be electrical; there should be an
 auxiliary  generator or a gasoline-driven pump in the event of power failure.
 The auxiliary  system  should be  periodically  tested according to a routine
 maintenance schedule.  There should be an alarm  system which operates inde-
 pendently  of the electrical system  and which is  tripped by any interruption
 in  water flow.   The alarm  itself should also be  checked regularly.

      3.  Cleaning—The distribution system should be designed for ease of
 cleaning.   If  the  entire system (including pumps) is installed in duplicate,
 then one system  can be shut down for cleaning while the other system is
 operational.   Periodic cleaning minimizes the problem of fouling.  Water
 allowed  to stagnate in the closed system will kill newly-settled larvae.  This
 system  is  then backflushed,  to  avoid discharge of anoxic water, sulfides, and
 organic  materials  into the microcosms.  To the extent possible, straight
 delivery lines should be used instead  of ones with many bends, (the straight
 lines also improve flow characteristics).  Open  bends should be used rather
 than tight elbows.  Each straight section of pipe should be directly acces-
 sible via  cleaning ports.

      4.  Flexibility—In order  to accommodate modifications towards differing
 experimental purposes, facility design should be flexible.  Light fiberglass
 tanks and  plastic  plumbing are  particularly  amenable to modification.  Suf-
 ficient  open working  area,  seawater delivery, drain systems, fresh water, and
 electrical power should be provided so that  facilities can be expanded and
 modified.   In  the  HIMB system,  groups  of three microcosms are served by a
 single inlet headbox.  The NUC  system  allows single or multiple microcosm
 service  by each  inlet headbox.   The latter arrangement is preferable.

     5.  Environmental modifications—Water  composition can be varied by -
 choice of  water  source (fresh water, bay water, well water), by manipulation
 before the water enters the head boxes, by modification in the head boxes (or
 inlet lines),land  by direct manipulation within  the tanks.  Natural, uncon-
 trolled  variations  in water quality should^-be minimized by choice of water
 intake sites.  Care should be ?fcaken to se%that  the microcosms are not shaded
by buildings or  other obstructions  at  anytime.

     6.  Plumbing  characteristics—Pipe^diameters to serve the microcosm
 system can be  calculated from an estimate of. the maximum height to which the
water must be delivered,'the ,total  length of.pipe in the distribution system,
 the  flow rates to be maintained, and the pump characteristics.  Engineering
handbooks  or plumbing experts should be consulted for detail, with some

                                       49

-------
 allowance made for fouling.   Our  facility  delivers  approximately  400  liters/
 minute,  to a maximum delivery height  of  5  meters  (an  aeration  tower).   The
 water flows through approximately 10  meters  of  pipe before  it  is  delivered  to
 the microcosms.   Minimum pipe diameters  in the  delivery  system are  1-1/4
 inches.   Valves  are not  used  to regulate water  flow.   Instead,  standpipes
 maintain constant water  levels and allow excess water to spill into drains.
 Flow rates are controlled by  head differences between constant-head reser-
 voirs and overflow pipes.

      Major considerations about the pump are:   1) that it be- capable  of
 delivering the required  flow  to the maximum  elevation in the system;  2) that
 it  be reliable under conditions of constant  operation; 3) and  that  all wetted
 surfaces be made of non-toxic materials.   It is desirable to have two primary
 pumps (in addition to the auxiliary).  One pump can be in operation while the
 other is down for routine maintenance or repair.

      We  have used centrifugal pumps.  Most small planktonic organisms en-
 trained  in the flow survive passage through  the pump,  although recent studies
 suggest  that larger and  more  delicate holoplankton  and meroplankton are
 killed.

      7.   Water characteristics—The water  used  for  most  experiments at HIMB
 is  drawn from a  depth of 2 meters  adjacent to a reef  slope in  Kaneohe Bay.
 The water is rich in plankton and  organic  detritus, so a fouling  community
 develops rapidly.   This  biological fouling necessitates  the double  seawater
 distribution system which  has been described.   Inlet  screens must be ex-    ;
 changed  and cleaned fortnightly.   A second water source  at HIMB is  a seawater
 well.  The salinity of that water  is  very  similar to  bay salinity.  There are
 other substantial chemical and biological  differences between  the two water
 sources.   Larvae and organic  detritus are  effectively  filtered from the well
 water.   The only organisms associated with the  well water are diatoms which
 appear to be introduced  as airborne contaminents.   The inorganic-nutrient
 level of the well water  is substantially higher than  that of the  bay  (Table ••
 8).   The well water  is anoxic and  rich in  hydrogen  sulfide.  It is  aerated by
 being flowed as  a "film" across rippled plastic roofing material  before being
 delivered to the microcosms.   That  treatment is sufficient to remove the H2S
 and bring the  02  concentration to  saturation.

      At  the  NUC  facility,  water is  drawn from a natural  sump at the seaward
 edge  of  the  reef flat.   This  sump,  protected from the force of the ocean waves
 by  the algal ridge,  allows-us  to draw water from beyond  the reef  crest.  In
 certain weather  conditions, water draining off  the reef flat flows into the
 sump  and  delivers nitrogen-rich water to the microcosm (see Henderson et at.,
 1976).  On rare  days during low spring tides when there is no surf action,
 the water in the sump is drawn down below  the intake level so that the
pumping ceases.  A standpipe  arrangement prevents the microcosms  from drain-
ing,  and  the pump is re-started manually.  The double pipe across the reef
flat  to  the sump was emplaced during a quiet-water spring low tide and then
secured with cloth bags which were filled,  in place, with concrete and
further anchored with reenforcing rods.   This emplacement has now survived
two seasons^of winter surf.  At the NUC facility a large amount of sand is
                                      50

-------
 suspended  in  the  surf  zone and pumped  into  the facility.  This problem is
 alleviated by delivering  the water  to  a 2,500-liter settling tank which
 allows  the sand to  settle out.  That sand is then cleaned out as needed.

 C.   Microcosm Aquaria. Associated  Devices  for Environmental Modification of
      Aquarium Characteristics                            ,

      The microcosm  tanks  used in our facilities are constructed of 3/8"
 thick sprayed  fiberglass.  The inner dimensions are 117 cm by 117 cm by 46 cm
 deep.  The total volume is 630 liters.  A horizontal flange approximately 5 cm
 wide  lends rigidity to the tanks.  Tank dimensions should be kept constant
 for inter-tank comparisons.  There are demonstrable microcosm community
 responses associated with surface to volume ratios and other "container
 effects."  The water depth in the tanks is usually kept near 35 to 40 cm
 (480-550 liters).   Water depth in the tanks and flow rate through the tanks
 are established by means of inlet and outlet head boxes.  The water depth in
 the head boxes is  fixed by means of adjustable standpipes (Figure A-2).
 Coarse control on the flow rate is controlled by the length of the standpipe.
 Screwing the standpipes up or down on threaded fittings provides fine control
 of flow. Such gravity control of flow rate is preferable to valves,  which
 rapidly become fouled or clogged.   With weekly cleaning of distribution lines,
 flow rates  can be  maintained to well within 5 percent of the desired rate by
 this system.   The  double standpipe arrangement also prevents the microcosms
 from draining in the event that seawater delivery stops.  Flow rate  through
 each microcosm is  usually maintained near 10 liters/minute,  so  the flushing
 time of each aquarium is about 50  minutes.   The drain on each outlet head box
 has a simple,  automatic (inverted-U) siphon so  that the head box fills,  starts
 the siphon, drains,  stops the  siphon,  and then repeats the  cycle.  Flow
 through the tanks  is precisely measurable by timing the filling  rate of  the
 head boxes.  Such  a  measurement system could be automated by installing
 electrodes  which registered  the time at two  known points (hence, volumes)  on
 the filling cycle.   We  have not found a preferable  inexpensive flow  meter
 which neither  interferes with  flow nor  remains  unaffected by fouling in the
 pipes.                            .-.,-,

     Water  flows into the  aquaria via a pipe (actually,  one  of two; one for
 each of  the duplicate distribution systems)  near  the center  of the aquaria.
 The water wells up,  mixes  through the tank,  and exits  through another (again,
 one of  two) pipe near one  corner of  the tank.  An earlier configuration of
 the inlet induced a  horizontal rotary flow from a point  inlet on one side of
 each tank.  This resulted  in a conspicuous gradient in the fouling community
 around the  tank sides.  Such a gradient induced by variable water motion
 precluded convenient sampling of an homogenous community; the gradient has
 been eliminated by the present design.

     Water composition can be varied by initial choice of water source (fresh
water, bay water, well water), by manipulation before the water enters the
 inlet head boxes, by mixing in the head boxes or the inlet lines, or by direct
additions to the tanks themselves.   Descriptions of how we alter water
composition follow.
                                     51

-------
AMBIENT (GRAVITY
Ft£0) INLET LINE

TEMPERATURE MIX
INLET LINE -

RESERVOIR  WAIN
(SETS LEVEL IN
 RESERVOIR)
INLET RESERVOIR
BOX

MANOMETER FORT-
UNES TO TANKS
W-COMWa
WATER SAMPLE
PEOCOCK
                                                      SECONDARY FEED LINE
                                                                                            (B)
                                                                              DRAIN BOX
                                                                          /"INDIVIDUAL CHAMBERS
                                                                               FOR EACH TANK)
                                                                              OUTLET STANOPIPE
                                                                              (SETS LEVEL IM TANK}
                                                                           OUT-COIHO WATER
                                                                           SAMPLE PEOCOCK
 Figure  A-2.   Arrangement  of  standplpes in inlet  (A)  and  outlet  (B)
                  head boxes.
                                                 52

-------
      Salinity is  decreased by  adding fresh water  to  the  head box at a  fixed
 rate.   The fresh  water is  delivered  by  gravity  feed  from a  large constant-
 head  reservoir, rather than being supplied directly  from the freshwater line
 which is  subject  to variable pressure and flow  rate.

      Dissolved inorganic materials (nutrients,  toxins, etc.), are added to
 the inlet boxes or  inlet lines by means of a peristaltic pump drawing  from a
 stock solution in distilled water.  By  using a  stock solution which is highly
 concentrated  relative  to the desired dissolved  material  concentration  in the
 microcosm (e.g. 1,000-fold), one  can avoid lowering  the  microcosm salinity
 with  the  additions.  Cations can  often  be added as chloride compounds, while
 anions are often  best  added as sodium compounds.  Because chloride and sodium
 are the two most  abundant  ions in seawater, use of such  compounds minimizes
 the effects of added materials other than those of specific interest.  Of
 course, salts of  sodium or chloride  are not always available or practical.
 Consideration of  available salts  and seawater chemistry  can suggest the most,
 viable choices.   For nutrient  additions under less precisely controlled
 conditions, we have also used  high-nutrient water from a seawater well (see
 Table  8).

      Temperature  is controlled by heating or chilling the inlet stream with a
 heat  exchanger system  (Jokiel  et  al.t manuscript) or by  placing glass resist-
 ance heaters  in the head boxes*   The former approach allows efficient and
 sophisticated control  but  involves extensive construction;  the latter is
 simple to operate,  but energy-intensive.

     Dissolved oxygen  concentration  is  lowered  by bubbling  nitrogen through
 the inlet head box.  It is  possible by this method to achieve a wide variety
 of  oxygen levels  simply by varying the  nitrogen bubbling rate and monitoring
 the water until the desired oxygen level is obtained.  If incoming water has
 low or variable oxygen level,  the values can be brought  to  near saturation
 by  cascading  the  water across  ripple panels.

 D.   Automatic Sampling and Data  Acquisition System

     1.   Automatic  water sampling system—The purpose of  this system is to
 control water flow  from various sources  to the  measurement  devices.   The unit
 is  under  the  control of a  data logger which advances solenoid values from one
 water  source  to the  next after each  measurement sequence.  Each channel is
 identified  by a feedback signal to a data logger  input channel.

     Each water source being monitored  is connected via a normally closed
 solenoid valve to a  common manifold  that directs water to, the test electrodes.
 Each valve  is actuated in  order by the sequencer.

     Various  electrical, pneumatic,  or hydraulic valves are available,  and
 different sequence  control  mechanisms can be used to meet specific needs.
 Multiple port distribution  valves  operated by stepping motors can hydraul-
 ically or pneumatically trigger the water sampling valves in sequence.   Elec-
 trically controlled  valves  are also  available.   New products are constantly
 coming onto the market  but  the following description of  the HIMB prototype
will serve as  an  example:

                                      53

-------
     Water  flow from the various  sources is controlled by 20 remote-
controlled  normally-closed one-inch valves.  The valves we have used  at  HIMB
are Toro  Company Model 202-08-04  units; these are made of plastic and offer
no resistance to flow when open.   The valves were modified by replacing  the
hard rubber seating gasket with soft 1/4 inch neoprene material to improve
seating.  These valves are hydraulic, operating off  freshwater  line
pressure; they are actuated by 24 volt current to small Automatic Switch
Company Model 8320A136 control valves.  The fresh water is isolated from the
seawater  stream.

     The  configuration of the HIMB water sampling system is shown in  Figure
A-3.  Only  one solenoid is activated at any given time by an electrical
signal from the data acquisition  system.  Water flushes the system for one
minute before data from the probes are recorded.  Specific flushing time
should consider maximum volume of the lines through  which the water flows,
minimum flow rate of the water, and maximum response time of any in-line
probes.   The manifold is designed to insure complete flushing; stagnant  areas
are eliminated by use of a loop.   Differences in flow over the probes that
occur due to variation in length  of hose can influence readings in the oxygen
electrode.   Constant flow over the probes at any water source pressure is
accomplished by using a double standpipe arrangement (Figure A-4).  Excess
water spills over the first standpipe; a constant difference in height and
thus constant flow over the probes is maintained.
             HOSES FROM MICROCOSM
              INLETS AND OUTLETS
                SEQUENCER
          SENDS SIGNAL TO OPEN ONE VALVE
          AT A TIME. IN SEQUENCE
                                FEEDBACK SIGNAL
        TO IDENTIFY
         SOURCE
4. LIGHT (INTEGRATED)

5 LIGHT (INSTANTANEOUS)
6. OPTIONAL

7. OPTIONAL	—

8. SOURCE 	•	
9. TIME	•
                                        DATA ACQUISITION
                                           SYSTEM
                                        CONTROLS SCAN RATE
                                        SELECTS CHANNEL
                                        ADVANCES SEOL-ENCER
                                        CONVERTS ANALOG TO
                                        DIGITAL SIGNAL
                                        INSERTS CHANNEL
                                        IDENTIFIER INTO DATA
                                        STREAM
                                           10. ADVANCE TO NEXT SOURCE
            Figure A-3.
Schematic diagram depicting the  HIMB water
sampling system.
                                        54

-------
            SURPLUS
           OVERFLOW
 WATER FROM
 MICROCOSMS
                         OXYGEN
                         PROBE
                       pH
                    ELECTRODE
                                                THERMJSTER
                                                   CONSTANT
                                                   OVERFLOW
   Figure A-4.
Enlargement of area of water sampling system (see Figure
A-3) which measures oxygen, salinity and temperature.
     2.  Automatic data acquisition system—Electronic data-loggers and
electronic measurement probes can be used to monitor water composition and
other variables in microcosms.  Many components of the systems utilized in
this program were designed and built locally by electronic technicians to
meet specific needs; since then, versatile commercial data acquisition
systems have become available.  The data acquisition system is simple in
concept.  The acquisition system signals the water sampling system to shunt
water from various sources past electronic sensors.  Outputs from the sensors
are recorded on magnetic tape and/or on a hard-copy printer.  A properly
designed system of this type is reliable and relatively easy to operate.  The
major problem usually centers on designing, building, and  debugging  the
unit.  The total system consists of measurement devices and a data logger
system.

     A variety of useful measurement devices are available.  The system
utilized in this program normally measures the following:  light, temperature,
dissolved oxygen, and pH.  Additional channels are available for input from
other instruments (turbidometer, fluorometer, specific ion electrode, con-
ductivity meter, etc.), as required.  Certain electrodes can interfere with
one another through the common ground of seawater (e.g. oxygen and pH
electrodes), so care must be exercised in matching instrument types.  Examples
of instruments used successfully with this, system are as follows:

         Temperature:  An electronic telethermometer (Yellow Springs
     Instruments Model 47) with epoxy coated probe (Model 402) was utilized;
     numerous other products would be equally suitable.  The probe is
     replaced with a 5 k-ohm variable resistor when intercalibrating the
     instrument with the data acquisition system.
                                     55

-------
         Oxygen:  Units with stable battery power sources  that are
      isolated  from line voltage  (e.g. Yellow  Springs Instrument
      Model  57) are unlikely to lose calibration or interfere with
      other  sensors.

         pH:   Most pH meters have  circuits designed to measure
      ungrounded  test solutions and will not function in a  grounded
      salt water  stream.  The Orion model 801A meter has an isolated
      test loop.  This meter is satisfactory when used with a
      combination electrode such  as the Beckman Model 3900.

         Light:  Instantaneous values of ambient light are  measured
      by a standard radiometric cell and/or are accumulated  value from
      a light integrator.  In the past few years commercially produced
      quantameters with integrators and recorder outputs have become
      available (Lambda Instruments) and provide a measure  that is
      superior  to radiometric values for biological studies.

      Our test  instruments are obsolete, because they produce analog outputs
which must be  converted to digital values in  the data logger.  Analog signals
are prone to transmission interference.  A superior system would utilize
digital output measurement devices exclusively.  Converters could be in-
stalled directly on analog output  instruments that could not be replaced with
digital output units.  We have been able to eliminate interference with our
analog signals by shielding the  wires and removing sources of electromagnetic
noise (high voltage transformers,  large motors, etc.), from  the immediate
vicinity of the data acquisition system.

      Up to 12 microcosms are usually monitored simultaneously in our system,
although fewer units are used for  some experiments.  The data acquisition
system is usually set to scan at one-minute intervals, but six different
scanning speeds are available on our logger.  Eighteen "water sources"
(twelve microcosm outlets, four  inlet head boxes, water from an aerated
saturation bath for Oa probe calibration, and one source in which the 02 meter
is grounded to zero as an event mark) are monitored in sequence.  On each
scan, the following variables are  recorded or cassette magnetic tape:  water
temperature, dissolved oxygen, pH, integrated total light,  "mark" (a feed-
back  voltage from the water sampling system used to identify each water
source) and "time" from an internal digital clock.  The data logger identifies
each  of the eight data channels on the tape by inserting a channel mark.  Use
of the YSI Model 57 oxygen meter has eliminated calibration drift encountered
with  the older YSI Model 54,  and use of the aerated calibration water source
on each cycle probably could be discontinued.   That calibration bath still
serves as an excellent check and reference mark in the data stream.

      Data cassettes are transcribed onto 9-track tape for use on an IBM 370
computer system.   The data are formatted into rows and columns and printed
for editing before data analysis.
                                     56

-------
                                    TECHNICAL REPORT DATA
                             (Please read Instructions on the reverse before completing)
 i. REPORT NO.
    EPA-600/3-79-061
                               2.
                3. RECIPIENT'S ACCESSION-NO.
 4. TITLE AND SUBTITLE
      Metabolic Responses of  Shallow Tropical  Benthic
      Microcosm Communities to  Perturbation
                                                             5. REPORT DATE
                                                               June 1979 issuing date
                6. PERFORMING ORGANIZATION CODE
 7. AUTHOR(S)
              S.  V.  Smith,

      P. L. Jokiel, G. S. Key,  and E. B. Guinther
                                                            8. PERFORMING ORGANIZATION REPORT NO
 9. PERFORMING ORGANIZATION NAME AND ADDRESS
      Hawaii  Institute of Marine Biology
      Kaneohe,  Hawaii 96744
                10. PROGRAM ELEMENT NO.

                  1BA608
                11. CONTRACT/GRANT NO.

                  R800906
 12. SPONSORING AGENCY NAME AND ADDRESS
       Environmental Research Laboratory
       Office of Research  and Development
       U.S.  Environmental  Protection Agency
      Narragansett,  Rhode Island 02882
                13. TYPE OF REPORT AND PERIOD COVERED
                     Final
                14. SPONSORING AGENCY CODE


                   EPA/600/05
 15. SUPPLEMENTARY NOTES
 16. ABSTRACT
           Benthos  communities simulating various aspects of coral reefs were
      established in  600-liter microcosm tanks.  These  communities were then
      subjected  to  various environmental perturbations,  including altered
      light regime, altered substratum type, salinity depression, elevated
      nutrient level,  and biological  manipulation.  The metabolic responses of the
      community  to  these perturbations were minortored,  primarily by analysis of
      dissolved  oxygen flux.  Light,  substratum type, and nutrient levels are
      resources  which  limit community metabolism.  From 35 to 22  /oo. Metabolism
      is not sensitive to salinty.  Metabolism is sensitive to biological
      manipulation.
                                KEY WORDS AND DOCUMENT ANALYSIS
                  DESCRIPTORS
                                               b.lDENTIFIERS/OPEN ENDED TERMS  C. COSATI Field/Group
     Marine Biology
     Benthos
     Metabolism
     Reefs
     a Ecosystems
   06 F
 8. DISTRIBUTION STATEMENT

     RELEASE TO PUBLIC
  19. SECURITY CLASS (ThisReport)
   UNCLASSIFIED
21. NO. OF PAGES

      67
                                               20. SECURITY CLASS (Thispage)

                                                UNCLASSIFIED
                             22. PRICE
EPA Form 2220-1 (9-73)
57
                                                                     OUSGPOi 1979-657-060/1675

-------