EPA-600/4-37/028
                                                         Hay 1988
SHORT-TERM METHODS FOR ESTIMATING THE CHROHIC TOXICITY OF EFFLUENTS

      AND RECEIVING WATERS TO MARINE AND ESTUARINE ORGANISMS
                              Edited
    Biologic    Method   Branch
Monitoring and Support uCatory
S.  Environmental Protection
        Cincifirwti,  Ohio
                                                    ,  J-
                                                    L' Robfnson
                                                    Cincinnati
                                                    Llncinn8U
              John Menkedick and Florence Kessler
                 Computer Sciences Corporation
                        Cincinnati,  Ohio
ENVIRONMENTAL MONITORING AND SUPPORT LABORATORY -
             OFFICE OF RESEARCH.AMU DEVELOPMENT
            U.  S.  ENVIRONMENTAL  PROTECTION AGENCY
                   CINCINNATI, OHIO  45268
                                  CINCINNATI

-------
NOTICE
recommendation for use
             endorsement 0

-------
                           FOREWORD


                          s?
 Measure the presence  and  concentration  of  physical
Concentrate, recover, and  identify enteric viruses, bacteria  «
other mcroorganis^s in water, wastewater, and municipal sludge

Measure the effects of pollution on freshwater, estuarine  and
marine organisms, including the phytoplankton, zooplaX
penphyton, macrophyton,  macroinvertebrates,  and fish

                                physfca1'

                           Thomas A.  Clark
                           Acting Director
                           Environmental Monitoring and
                           Support Laboratory - Cincinnati

-------
                                   PREFACE

    This manual is the second Agency methods manual for estimating the
chronic toxicity of effluents and receiving waters.  The draft was
reviewed by the Bioassay Subcommittee of the EMSL-Cincinnati Biological
Advisory Committee, USEPA headquarters and regional staff, other Federal
agencies,  state and interstate water pollution control programs,
environmental protection groups, trade associations, major industries,
consulting firms,  academic institutions engaged in aquatic toxicology
research,  and other interested parties in the private sector.
                                        A!LBJMMJJ1EA MEMBERS     ff      .;

    William Peltier,  Subcommittee Chairman,                               r
    Environmental  Services  Division,  Region  4
    Peter  Nolan,  Environmental  Services  Division,  Region  1
    Roland Hemmett, Environmental  Services Division,  Region  2
    Ronald Preston, Environmental  Services Division,  Region  3       C
    Charles Steiner,  Environmental Services  Division, Region 5      -;
    Terry  Hollister,  Environmental Services  Division, Region 6
    Bruce  Littell, Environmental Services Division, Region 7
    Loys Parrish, Environmental Services Division, Region 8
   Hilton Tunzi, Environmental Services Division, Region 9
   Joseph Cummins, Environmental Services Division, Region 10
   Bruce Binkley, National Enforcement Investigations Center, Denver
   Wesley Kinney, Environmental Monitoring Systems Laboratory --Las Vegas
   Steven Schimmel,  Environmental Research Laboratory - Narragansett
   Douglas Middaugh,  Environmental Research  Laboratory - Gulf Breeze
   Donald Mount,  Environmental Research  Laboratory - Duluth
   Alan Nebeker,  Environmental Research  Laboratory - Corvallis
   Margarete Heber,  Program Development  Branch,  Permits Division,
     Office of Water  Enforcement  and  Permits
   Edward  Bender,  Compliance Branch,  Enforcement Division, Office  of
     Water Enforcement and  Permits
   James  Plafkin,  Monitoring Branch,  Monitoring  and  Data  Support
     Division,  Office of  Water  Regulations and Standards
   Christopher  Zarba,  Criteria  and Standards Division,  Office of Water
     Regulations  and  Standards
   David Coppage,  Environmental Effects Branch,  Health  and Environmental
     Review Division,  Office of Toxic Substances
   Daniel  Rieder, Office  of Pesticides Programs
                            Cornelius I. Weber, Ph.D.
                            Chairman, Biological Advisory Comnittee
                            Chief, Biological Methods Branch
                            Environmental Monitoring and Support
                            Laboratory - Cincinnati
                                   IV

-------
  ABSTRACT
 for
 five species:  the  sheepshead
 silverside. Menidia bw
Also  included are^jUTdiTmes^n
taciimes and equipment  J-'n •
data analysis, report
Listings of computer programs
are provided in  the Appendix.
the
                                -thod,
                   •  rece'vin9 waters to
                            the inland
                            the sea
                                    ,
                       ling and holdino
                       g L handling '
                     and Probit Analysis

-------
                                   CONTENTS

 Foreword 	
 Preface	     	    1"
 Abstract	.* i  !!!!!!	     iv
 Figures  ......           '          	      v
 Tables	i i ."."•!"!!!!	    Vi.l
 Acknowledgments  ...       	*  .  .  .    vii
                          	*	viii
       1.  Introduction  	

       a'!  Heathen™ Sal^f f?r.E!t!"?t:n«  a™*  ™«'*   :  :  :      *
       4.  Quality Assurance   	'    	      ^
       5.  Facilities and  Equipment   ..!!!!  	    ]9
       6.  Test  Organisms	              	     '£
       7.  Dilution Water	\	     °

       I*  rhrnnfrV"-  K^™1"^ w*^r Sampling and  Samp lY Handing    20
      in*  ShronJCnToxiclty  Test End Points and Data Analysis   .        ?3
      10.  Report  Preparation   ....                                  ^i
      11.  Sheepshead Minnow  (£^rjnp^ ^aM^t^s) Larval'siriiCal
           and Growth  Test	       	~                     „,
      12.  Sheepshead Minnow  (C^npdpji varjia^is) Embryo-larval '
           Survival and Teratogenicit7 test  . .                      8fi
      13.  Inland Silverside  (HenidJa beryl!ina) Larval Survival * *
           and Growth Test	~T   —                        1?1
      14. Mysid (Midoris bahia)  Survival/Growth/and	
           Fecundity Test	
      15. Sea Urchin (Arbacia Runctujataj Fertilization Test*  \\\
      lo. Algal  (Champja HnyJd£TRepr5duction Test ........

Selected References   	
Appendix	.'!.*."	    Q

     A. Independence,  Randomization, and Outliers   .  .                320
     B. Validating Normality and Homogeneity of  Variance       " *
          Assumptions	
     C. Dunnett's Procedure	
     D. Bonferroni's T-test	
     E. Steel's Many-one  Rank  Test   . .  . .'	" "
     F. Wilcoxon  Rank  Sum Test	         	
     G. Probit Analysis	     	
                                    VI

-------
  Number
       Control  chart
                                    FIGURES
                                                                      15
                                                                      30
Number
                                   TABLES

                               V17

-------
                                  ACKNOWLEDGMENTS

   Protection Agency, Washington, D.C.;  Peltier  T H   and ^™"f tal
   1985, Methods for Measuring the Acute Tox  citv of fffl^l^ I" r*^' eds"

   c'ScKl W?' .™«^"£™$ anV       \ b°o ^ f!"







       a                 'U      nV
                 -85                               z        r
  Conduct  and  Interpretation of complex Eff uent^ic  tTle ts It*"

      aeMa,nnre SHeS' Emlr°menM R*s^ Laboratory  - Narraaansett
              S.  Env^onmental Protection Agency,  Narragansett.Kfsl.nd
                                         y



                                                                 6-
^Current  address: Program Development Branch,  Permits
 Office of Water Enforcement and Pprmitc  ,7 c   r   -          ,
 Agency,  Washington, DC              lt$>  U* S'  Erivi^menta1 Protection

-------


   the methods sections and the Appendix
                                                                sections in
                       uassn               '  dese™es special
manual.  Many helpful  uggest?on  for IHIJf d in !he Preparation  of  the
Provided by members of th" Bioassay Subc^ttee?"    ^  ^^  W6re also
                               n                                 >
Washington,  DC;  R.  L.  Casoe   U  s   Fn>   Envlro™e"tal Protection Agency,
                          *i r

                              !    «
     T. J. Hall, National Council of the Pa^J ?°nlt?nr,     .,       s
  Improvement, Inc., Anacortes, WA; S  SaU  P|UAB? U?try f?r  Al> and  Stream
  D. Hansen, Environmental Research Laboratorv  !J  \  "P" -Nashv111e.  TN;
  Agency, Narragansett,  RI; w.  D.  Holleraan   Al%h;J'nErlVlrqnmenta1 Protection
  Environmental Management  Montgomery  ™ "'.   L ^nf8**"16^ °f
 Department of Environmental Protection   T«;*h  K^nd°efei". Jr., New Jersey
 Department of Environmenta  Conservation  A  ban;  NV- N ^ K,U2J8' New Y^k
 Roads  Sanitation District, Virginia Beach  VA  {' ^Y; N' ,E'  LeBlanc, Hampton
 Protection Agency, Gulf  Breeze, FL  A  J 'N^II' t0we{ U'  S'  E^ironmental
 of Environmental Conservation! Albany" m-  M  H '^ J0rk ,State  Department
 Institute  of  Marine Science,  Gloucester Poin^  M Ro.bei"ts,  Jr.,  Virginia
 Department of Environmental Regulation  T»n»h«  J    .,T> Ross>  n°n: W'         ,
                                                       5' . Environmental
                                                      Environmental

                                                    Pe°7eultl '""flute,

                                                              Ed1son' -"J

-------
                                    SECTION 1

                                   INTRODUCTION                      ^^-

  1.1  This manual  is  intended to  serve as  a companion to the freshwater and
  marine acute toxicity test manual  (EPA-600/4-85-013) and the freshwater
  chronic toxicity  test manual (EPA-600/4-85-014) published earlier  by the
  Environmental Monitoring and Support Laboratory - Cincinnati
  ^t^/Uw???10^ "fu 1n the Nat1onal Pollutant Discharge Elimination
  System (NPDES).   These three toxicity test manuals have been prepared to
  assist the Agency in meeting the goals of the Federal Water Pollution
  Control Act Amendments of 1977, the Clean Water Act (CWA) of 1977
  (PL 95-217 , and  the Water Quality Act of 1987, which were enacted to
  restore and maintain the chemical, physical, and biological integrity of the
•Nation's waters {Section 101[a]), and contained specific or implied
  requirements for the collection of biomonitoring data in at least  15
  sections.

  1.2  The Declaration of Goals and Policy,  Section  101(a}(3),  in these two
  laws,  states that "it is the national goal that the discharge  of toxic
 pollutants in toxic  amounts  be  prohibited."   To achieve the goals of this
 legislation,  extensive  effluent toxicity screening programs were conducted
 ?a«ng^?f.1970?n&,the re9ions and states.   Acute toxicity tests (USEPA,
 1975;  Peltier,  1978}  were used  to measure  effluent toxicity and to estimate
 the safe concentration  of toxic effluents  in  receiving  waters.   However,  for
 those  effluents  that  were not sufficiently toxic to  cause mortality in  acute
 tone-  to four-day) tests,  short-term inexpensive methods were  not available
 to  detect  the more subtle, low-level,  long-term, adverse effects of
 effluents  on  aquatic  organisms,  such as  reduction  in growth and
 reproduction, and  occurrence,of terata.  Fortunately, rapid developments  in
 toxicity test methodology during  the past  five years have resulted  in the
 availability of  several methods  that permit detection of the low-level
 adverse effects  (chronic  toxicity)  of effluents in nine days or  less.  '

 1.3  As a result of the increased awareness of the value of effluent
 !in^lty t?st data for toxics control in the water quality program  and the
 NPDES permit program, and the recent availability of short-term  chronic
 toxicity test methods, the U. S. Environmental Protection Agency (USEPA)
 issued a national  policy statement entitled, "Policy for the Development of
Water Quality-Based Permit Limitations for Toxic Pollutants," in  the Federal
Register Vol. 49,  No. 48, Friday, March 9, 1984.  A technical support
document on the use of effluent and receiving water toxicity data also has
been prepared by the Office of Water Enforcement and Permits (OWEP) and the
Office of Water Regulations and Standards  (OWRS) to provide additional
guidance on the implementation of the biomonitoring policy (USEPA,  1985).

1.4  This policy recommends the  use of toxicity data to assess  and control
the discharge of toxic substances to the Nation's waters through the NPDES
permits program under Section 402 of the Clean Water Act.   The  policy states
that 'biological  testing of effluents is an important aspect of the
                                       1

-------
 water  quality-based  approach  for  controlling  toxic  pollutants.   Effluent
 toxicity data,  in conjunction with  other  data,  can  be  used  to establish
 control priorities,  assess  compliance with  State water quality  standards,
 and  set permit  limitations  to achieve those standards."  All states  have
 water  quality standards which include narrative statements  prohibiting the
 discharge of toxic materials  in toxic amounts.

 1.5  There  is a need for short-term toxicity  tests,  simillar to  those
 developed for the freshwater  organisms, to  evaluate  the toxicity of
 effluents discharged to estuarine and coastal marine waters under the NPDES
 permit program.  Methods are  presented in this  manual  for five  species from
 four phylogenetic groups.   Five of  the six  methods were developed and
 extensively field tested by Environmental Research Laboratory -
 Narragansett (ERL-N) over the last  three years.  The methods vary in
 duration from two hours to  nine days.  It is  anticipated that during the
 next two years, methods will  be developed for plant and animal species from
 other phylogenetic groups and other geographic  areas,  including  the Pacific
 Coast.

 1.6  The five species for which toxicity test methods  are provided are:
 the sheepshead minnow, Cyprinodon yariegatus; the inland silverside,
Menidia beryl!ina; the mysid, Mysidopsis bahia; the sea urchin,   Arbacia
 punctulata; and the red, macroalga, Champia parvula.

 1.6.1  The tests included in this document  are  based on the following
methods:

    1. "Guidance manual for conducting complex  effluent and receiving
       water larval  fish growth/survival  studies with the sheepshead
       minnow (Cyprinodon variegatus),"  by Melissa M. Hughes,  Margarete
       A.  Heber, Steven C.  Schiramel and  Walter 0.  Berry,  1987,
       Contribution No. X104, Environmental  Research Laboratory, U.  S.
       Environmental  Protection  Agency,  Narragansett, Rhode Island.

    2.  "Guidance manual for rapid chronic toxicity test on effluents and
       receiving waters with larval inland silversides (Menidia
       beryl 1ina?,"  by Margarete A. Heber, Melissa M. Hughes,  Steven C.
       Schimmel, and  David  Bengtson, 1987, Contribution No.  792,
       Environmental.Research Laboratory,  U. S.  Environmental  Protection
       Agency,  Narragansett, Rhode Island.

    3.  "Guidance manual for conducting seven-day,  mysid
       survival/growth/reproduction study using  the estuarine  ruysid,
       Mysidopsis  bahia,"  by Suzanne M.  Lussier, Anne Kuhn,  and  John
       Sewal1,T987, Contribution No. X106, Environmental  Research
       Laboratory,  U. S.  Environmental Protection  Agency,  Narragansett,
       Rhode Island.

-------
     4.  "Guidance manual for conducting sperm cell  tests with the sea urchin,
        Arbacia punctulata,  for use in testing complex effluents," by Diane E.
        Nacci,  Raymond Walsh,  and Eugene Jackim,  1987, Contribution No.  X105,
        Environmental  Research Laboratory,  U.  S.  Environmental  Protection
        Agency, Narragansett,  Rhode Island.

     5.  "Guidance manual for conducting sexual  reproduction  tests with the
        marine  macroalga Champia parvula for  use  in testing  complex
        effluents,"  by Glenn B.  Thursby and Richard L.  Steele,  1987,
        Contribution No. X103,  Environmental  Research  Laboratory,  U*  S.
        Environmental  Protection Agency,  Narragansett,  Rhode Island.

     6.  A  nine-day,  sheepshead  minnow  (Cyprinodon variegatus),  static-renewal,
        embryo-larval  survival  and  teratogemcity test,  developed  by  Terry
        Hollister, USEPA, Region 6, Houston, Texas.

 1.6.2   Four of the  methods  incorporate  the chronic  end  points  of  growth  or
 reproduction (or  both)  in addition to  lethality.   The  sheepshead  minnow  9-day
 embryo-larval  survival  and  teratogenicity test incorporates  teratogenic
 effects in addition to  lethality.  The  sea urchin  sperm cell test  uses
 fertilization as  an end point and has the advantage of an extremely  short
 exposure period  (1  h  and 20 min).

 1.6.3   These methods  were developed to provide the most favorable
 cost-benefit relationship possible, and are intended for use in effluent
 toxicity tests performed on-site, where time is very costly, and for toxicity
 tests with effluent samples shipped off-site to distant laboratories,
 requiring that the volume of waste shipped be kept to a minimum.

 1.7  EMSL-Cincinnati  has incorporated the short-term chronic and sub-chronic
tests into this manual for use by regulatory agencies involved in biological
monitoring of wastewater under the NPDES program.  Authority for promulgating
test procedures for the analysis of pollutants is contained in Section 304{h)
of the CWA.

1.8  The manual was prepared in the established EMSL-Cincinnati format (Kopp,
1983)*                                       :;,                            KH

-------
             .....                    SECTION 2

                SHORT-TERM METHODS FOR ESTIMATING CHRONIC TOXICITY



                                                                    =.










 over relat7vely short  exposure periods (two-to-four days)   9       '     '

                                                 »"
 laboratory  life-cycle tests may not accurately  estimate the "sale"
2.5  McKim (1977)  evaluated the data from 56 full life-cycle  tests  V

which used the  fathead minnow, and concluded that the embryo-larval and

-------
  2.6  Macek and Sleight (1977)  found that exposure of critical  life-staqes of
  fish to toxicants provides estimates of chronically safe concentrations
  remarkably similar to those derived from full  life-cycle toxicity tests  and
  reported that  "for a great majority of toxicants,  the conception wh  ch
  will not be acutely toxic  to the  most sensitive  life stages  is  the
  chronically safe  concentration for  fish,  and that  the most sensitive life
  stages  are the embryos and fry."  Critical  life-stage exposure  was
  considered to  be  exposure  of the  embryos  during  most,  preferably  all, of  the
  embryogenic  (incubation) period,  and  exposure  of the fry for 30 days
  post-hatch for warm water  fish  with embryogenic  periods  ranging from
  one-to-fourteen days,  and  for  60  days  post-hatch for fish with  longer
  embryogenic periods.   They concluded  that in the majority of cases?  the
  maximum  acceptable  toxicant  concentration (MATC) could be estimated  from  the

  Says Jos?-hatch!Ure         embry°S dUM'n9 1ncubat1on> a"d the larvae for  30

  2.7  Because of the  high cost of full  life-cycle fish toxicity  tests and the
  emerging concensus that the ELS test data usually would be adequate for
 estimating chronically safe concentrations,  there was a rapid shift by
 aquatic toxicologists to 30- to 90-day ELS toxicity tests for estimating
 chronically safe concentrations in the late  1970s.   In 1980,  USEPA adopted
 the po icy that ELS test data could  be used  in  establishing water quality
 criteria if data from full  life-cycle tests  were not available   (USEPA,,


 2.8  Published  reports of the results of ELS tests  indicate that the
 relative sensitivity of growth  and survival  as  end  points may be species
 dependent  toxicant dependent,  or  both.  Ward and Parrish (1980) examined
 the literature  on  ELS tests that used  embryos and juveniles of the
 sheepshead minnow  (Cyprinodon variegatus). and  found that growth was not a
 statistically sensitive indicator  oTtoxicity in  16 of 18 tests    Thev
 Sh$tKd ti-t-th! ,ELS te$tS be shortened  to 14 days posthatch and that
 growth be eliminated as an  indicator of toxic effects.

 2.9  In  a review of  the literature on  173  fish  full  life-cycle and  ELS tests
 performed to determine the  chronically  safe  concentrations of a  wide  variety
 of  toxicants, such as metals, pesticides,  organics,  inorganics,  detergents
 and complex effluents,  Woltering (1984) found that  at  the  lowest effect
 concentration,  significant  reductions were observed  in fry survival  in 57%
 fry growth  in 36%, and  egg  hatchability in 19%  of the tests.  He a  so found
 that  fry  survival and  growth were very  often equally  sensitive,   and

 T± ns 5aVhe ?[°Wth resp°?se C0u1d be deleted fr™ routine application
 of the ELS tests.  The  net  result would be a significant reduction  in the
 duration and cost of  screening tests  with no appreciable  impact  on
 estimating MATCs for chemical hazard  assessments.   Benoit, et  al     (1982)
±!i7;  'K e!rly "Je-stage tests  with four  organic chemicals,  found larval
growth to be the most significant measure of  effect, and survival to be
equally or less  sensitive than growth.

-------
   2.10   Efforts  to  further reduce the length  of partial  life-cycle toxicity
   tests  for  fish without  compromising their predictive value have resulted  in
   the development of  an eight-day,  embryo-larval  survival  and teratogenicity
   test for fish  and and other  aquatic vertebrates (Birge and Black,  1981;
   Birge  et al.,  1985), and  a seven-day larval  survival and growth  test  (Mount
   et al., 1984;  Norberg and Mount,  1985).

   2.11   The similarity of estimates of chronically safe concentrations of
   toxicants derived fromshort-term,  embryo-larval survival  and teratogenicity
   test to those derived from full life-cycle tests has been  demonstrated by
   Birge et al. (1981), Birge and Cassidy (1983), and Birge et al.  (1985).

  2.12  Use of a seven-day, fathead minnow larval survival  and growth test was
  first  proposed by Norberg and Mount at the 1983 annual meeting of the
  Tnoo?  y t°r Environmenta1 Toxicology and Chemistry (Norberg and Mount
  1983).   This test  was subsequently used by Mount and associates in field
  demonstrations  at  Lima,  Ohio  (Mount, et al.,  1984),  and at many other
  locations.   Growth was frequently found to be more  sensitive than survival
  in  determining  the effect  of  complex effluents.

  2.13  Norberg and Mount  (1985)  performed  three single toxicant  fathead
  minnow  larval growth  tests with  zinc,  copper,  and DURSBAN*, using dilution
  water from Lake  Superior.  The  results  were comparable to,  and  had
  confidence  intervals  that overlapped with, chronic values reported  in the
  literature for  both ELS and full  life-cycle tests.

  2.14  Hughes et  al. (1987) and Heber, et. al., (1987) adapted the fathead
 minnow  larval growth and survival test  for use with, the sheepshead minnow
 and the inland silverside, respectively.  When daily renewal 7-day
 sheepshead minnow larval  growth and  survival  tests and 28-day ELS tests were
 performed with industrial and municipal effluents, growth  was more sensitive
 than^survival in seven out of 12 larval growth and survival tests, equally
 sensitive in four tests,  and  less sensitive in only one test.  In four
 cases,  the  ELS test may ha've  been three to 10 times  more  sensitive to
 ?rflMenn? than ^he  i?rval  growth and survival  test.   In tests using copper,
 the  No  Observable Effect  Concentrations (NOECs) were  the same for both types
 of test, and  growth was the most sensitive end point  for both.   In a
 four-laboratory  comparison, six  of seven tests produced  identical  NOECs  for
 survival and  growth (Schimmel,  1987).  Data indicate  that  the inland
 silverside  is at least equally sensitive or more  sensitive to effluents  and
 single compounds than  the  sheepshead  minnow, and  can  be tested over  a  wider
 salinity range,  5-30 o/00  (Schimmel,  1987).

 2.15  Lussier et  al.  (1985a) determined  that survival  and  growth  are often
 as sensitive  as  reproduction in  28-day life-cycle tests with Mysidopsis
 bahia.        .                                                -*	e—

 2.16  Nacci et al., 1985 compared the  results from the sea  urchin
 fertilization test, using organic compounds, with results  from acute
 toxicity tests using the freshwater organisms, fathead minnows and Daphnia
ma^na.  The test was also compared to acute toxicity tests  using Menidia
rcemdia and Mysidopsis bahia and five metals.   For six of the eight organic

-------
compounds, the results of the fertilization test and the acute toxicity test
correlated well (r* = 0.85).  However, the results of the fertilization
test with the five metals did not correlate well with the results from the
acute tests.

2.17  Thursby and Steele (1987) evaluated two industrial effluents
containing heavy metals, five industrial effluents containing organic
chemicals (including dyes and pesticides), and  15 domestic wastewaters using
the two-day Champia parvula sexual reproduction test.  Nine single compounds
were used to compare the effects on sexual reproduction using a two-week
exposure and a two-day exposure.  For six of the nine compounds tested, the
chronic values were the same for both tests.

2.18  The use of short-term toxicity tests including subchronic and chronic
tests in the NPDES Program is especially attractive because they provide a
more direct estimate of the safe concentrations of effluents  in receiving
waters than was provided by acute toxicity tests, at an only  slightly
increased level of effort, compared to the fish full life-cycle chronic and
28-day ELS tests and the 28-day mysid life-cycle test.

-------
                                   SECTION 3

                              HEALTH AND  SAFETY 1
3.1  GENERAL PRECAUTIONS
3.1.1  Collection and use of effluents in toxicity tests may involve
significant risks to personal safety and health.  Personnel collecting
effluent samples and conducting toxicity tests should take all safety
precautions necessary for the prevention of bodily injury and illness which
might result from ingestion or invasion of infectious agents, inhalation or
absorption of corrosive or toxic substances through skin contact, and.
asphyxiation due to a lack of oxygen or the presence of noxious gases.

3.1.2  Prior to sample collection and laboratory work, personnel should
determine that all necessary safety equipment and materials have been
obtained and are in good condition.

3.2  SAFETY EQUIPMENT

3.2.1  Personal Safety Gear

    Personnel should use safety equipment, as required, such as rubber
aprons, laboratory coats, respirators, gloves, safety glasses, hard hats,
and safety shoes.  Plastic netting on glass beakers, flasks and other
glassware minimizes breakage and subsequent shattering of the glass.

3.2.2  Laboratory Safety Equipment

    Each laboratory (including mobile laboratories) should be provided with
safety equipment such as first aid kits,  fire extinguishers, fire blankets,
emergency showers, and eye fountains.

3.3  GENERAL LABORATORY AND FIELD OPERATIONS

3.3.1.  Work with effluents should be performed in compliance with accepted
rules pertaining to the handling of hazardous materials (see safety manuals
listed in Paragraph 3.5).  It is recommended that personnel collecting
samples and performing toxicity tests should not work alone.

3.3.2.  Because the chemical composition  of effluents is usually only
poorly known, they should be considered as potential health hazards,, and
exposure to them should be minimized.

3.3.3.  It is advisable to cleanse exposed parts of the body immediately
after collecting effluent samples.

3.3.4.  All containers should be adequately labeled to indicate their
contents.
^Adapted from: Peltier and Weber (1985).

                                      8

-------
  3.4   DISEASE  PREVENTION
3.5  SAFETY MANUALS






samples «ndfconduct?nJdtSxS?aC1CS When
                                                                  a"d
3.6  WASTE DISPOSAL                                                        i

-------
                                   SECTION 4

                               QUALITY ASSURANCE!
 4.1   INTRODUCTION
 4.1.1   Quality Assurance  (QA)  practices  for effluent  toxicity tests  consist
 of  all  aspects of  the  test  that  affect data quality,  such  as:  (1)  effluent
 sampling  and  handling;  (2)  the source and  condition of  the test organisms;
 (3)  condition of equipment;  (4)  test conditions;  (5)  instrument calibration;
 (6)  replication; (7) use  of  reference toxicants;  (8)  record keeping;  and  (9)
 data evaluation.   For  general  guidance on  good  laboratory  practices  related
 to  toxicity testing, see:   FDA,  1978; USEPA,  1979d, 1980b,  and 1980c;  and
 DeWoskin,  1984.

 4.2  EFFLUENT AND  RECEIVING  WATER  SAMPLING AND  HANDLING

 4.2.1   Effluent samples collected  for on-site and off-site  testing must be
 preserved  as  described  in Section  8, Effluent and Receiving Water Sampling
 and  Sample Handling.

 4.3  TEST  ORGANISMS

 4.3.1   The test organisms used in  the procedures described  in  this manual
 are  the sheepshead minnow (Cyprinodon variegatus); the inland  silverside
 (Henidia beryl!Ina); the mysid (MysTdopsis  bahia); the sea  urchin (Arbacia
 punctulata) and the macroalga  (Champia paryula).  The organisms  used should
 be disease-free, and should  be positively  Identified to species.  Suitable
 taxonomic keys are cited in  the toxicity test methods.

 4.4  FACILITIES, EQUIPMENT, AND TEST CHAMBERS

 4.4.1  Laboratory and bioassay temperature control equipment must be
 adequate to maintain recommended test water temperatures.  Recommended
 materials must be used in the fabrication of the test equipment which comes
 in contact with the effluent (see Section 5, Facilities and Equipment).

4.5  ANALYTICAL METHODS

4.5.1  Routine chemical and physical analyses must include established
quality assurance practices outlined in  Agency methods manuals (USEPA,
 1979a,b).

4.6  CALIBRATION AND STANDARDIZATION

4.6.1  Instruments  used for routine measurements of chemical and physical
parameters such as  pH,  DO, temperature,  conductivity,  alkalinity, and
salinity,  must be calibrated and  standardized according to instrument
1 Adapted from: Peltier (1978), Peltier and Weber (1985),
 and USEPA (1979a).

                                      10

-------
Method 130.2, USEPA  979b)

4.7  DILUTION WATER
                                                                  be
                                           in the *P«:tftc EPA method  (see EPA
            used  should be appropriate to
            l  constraints, as^cussed
   4.8   TEST CONDITIONS
            of the testad

  4.9  TEST  ACCEPTABILITY
                                                          within
                                              o- - <** tests are
 test requires control egg fertili^? L  f    , °J 9reate'"-  The sea urchin
 greater than 90% fertfllK  ^  resul^n'J^?1' e;ceedi"9 7M-  ^ver,
         test is acceptable if survival   , im?"kl!!9 Joxic resP°"ses.   ThT
         pS ?er P'"t  should equal or exceed ?o' "?? !hhe mean  number  "f






formalin solution.   The mean mv^H JI the.1ar^e are preserved in  a 4%
                                    n

-------
4.10  PRECISION

4.10.1 The ability of the  laboratory personnel to obtain consistent, precise
results must be demonstrated with reference toxicants before they attempt to
measure effluent toxicity.  The single  laboratory precision of each type of
test to be used in a laboratory should  be determined by performing at least
five or more tests with a  reference toxicant.

4.10.2  In cases where the test data are used in the Probit Analysis (see
Section 9), precision can  be described  by the mean, standard deviation, and
relative standard deviation (percent coefficient of variation, or CV) of the
calculated end points from the replicated tests.  However, in cases where
the results are reported in terms of the NOEC and Lowest-Observed-Effect
Concentration (LOEC) (see  Section 9)s precision can only be described by
listing the NOEC-LOEC interval for each test.  In this case, it is not
possible to express precision in terms  of a commonly used statistic.  For
instance, when all tests of the same toxicant yield the same NOEC-LOEC
interval, maximum precision has been attained.  However, the "true" no
effect concentration could fall anywhere within the interval, NOEC + (LOEC
minus NOEC).

4.10.3  It should be noted here that the dilution factor selected for a test
determines the width of the NOEC-LOEC interval and the inherent maximum
precision of the test.   As the absolute value of the dilution factor
decreases, the width of the NOEC-LOEC interval increases, and the inherent
maximum precision of the test decreases.  When a dilution factor of 0.3 is
used, the NOEC could be considered to have a relative uncertainty as high as
+ 300%.   With a dilution factor of 0.5, the NOEC could be considered to have
a relative variability of + 100%.   Other factors which can affect test
precision include: test organism age, condition, and sensitivity;
temperature control; and feeding.

4.11 REPLICATION AND TEST SENSITIVITY

4.11.1  The sensitivity of the tests will  depend in part on the number of
replicates, the probability level  selected, and the type of statistical
analysis.  The minimum recommended number of replicates varies with the test
and the  statistical  method used, and is discussed in Section 9 and in each
method.   The sensitivity of the test should increase as the number of
replicates is increased.

4.12  QUALITY OF TEST ORGANISMS

4.12.1  If the laboratory does not have an ongoing test organism culturing
program and obtains  the test organisms from an outside source,  the
sensitivity of each  batch of test  organisms must be evaluated with a
reference toxicant in a short-term chronic toxicity test performed
concurrently with the effluent and/or receiving water toxicity tests.  If
the laboratory maintains breeding  cultures, the sensitivity of the offspring
should be determined in a toxicity test performed with a reference toxicant
at least once each month.  If preferred, this reference toxicant test may be
performed concurrently with an effluent toxicity test.

                                      12

-------
    ™c,2*r-  ™ree Deference tQxicants are presently available  from
    EMSL-CincnnnatT to establish the precision and validity nf fnlv,-^ A +
    generated by biornonitoring laborat'orie ^   sodiurdodecy] Llfate  fs^f ," „
    cadrmum chloride  CdCl2),  and copper sulfate  /r^sn,?    Thf.       (  S) and
    toxicants may be obtained  by coKi  g  th q  ^  itftssu?^  n
    Env7ronmental Monitoring and Support               ?ss"r™e

                       '            '
   values for the  reference

   4.13   FOOD QUALITY
  4.14  CONTROL CHARTS






          1             eChn1                      "'         the
 control limits (+2S  are re-Mlcul.:ih   ^     ^ Upper and 1ower
4.15  RECORD KEEPING
                                     13

-------
age, date of receipt, and other pertinent information relating to their
history and health, and information on the calibration of equipment and
instruments, test conditions employed, and test results.  Annotations should
be made on a real-time basis to prevent the loss of information
                                    14

-------
   UJ
   o.
             UPPER CONTROL LIMIT
                 CENTRALTENDENCY
            LOWER CONTROL LIMIT
    o
    UD
    O
             UPPER CONTROL LIMIT|X+ 2S)
                 CENTRALTENDENCY
             LOWER CONTROL LIMIT{X - 2S)
              1
                               15
                                       20
       0       5        10
      TOXICITY TEST WITH REFERENCE TOXICANTS      f|

  Figure  1. Control  charts: (A) hypothesis testing results;
           (B) Probit Analysis data.
               X -
                   n
Where:     Xj = Successive toxicity values from toxicity tests
          n_ = Number of tests.
          X = Mean toxicity value.
          S = Standard deviation.

-------
                                     SECTION 5

                             FACILITIES AND EQUIPMENT^
  5.1  GENERAL REQUIREMENTS

  5.1.1  Effluent toxicity tests may be performed in a fixed or mobile
  n±°n±cy" rF?'imie* «ust Include facilities for holding and acclimating
  organisms.   Culturing facilities for test organisms may be desirable in
  fixed laboratories which perform large numbers of tests.  Temperature
  control  can be achieved using circulating water baths,  heat exchangers, or
  environmental  chambers.  Water used for rearing,  holding,  acclimating,  and
  testing  organisms  may be natural sea water or hypersaline  brine derived from
  natural  sea water,  or water  made up from artificial  sea  salts  when
  specifically recommended in  the  method.   Air  used  for aeration must  be  free
  of  oil and  toxic^vapors.   Test facilities  must be  well ventilated and free
  of  fumes.   Organisms  should  be shielded  from  external disturbances.

  5.1.2  Materials used  for  exposure  chambers,  tubing, etc., which come in
  contact with the effluent, should be chosen carefully.   Tempered glass  and
  perfluorocarbon plastics  (TEFLON*)  should  be  used whenever possible  to
 minimize sorption and  leaching of toxic  substances.  These materials  may be
 reused following decontamination.  Plastics such as polyethylene
 polypropylene,  polyvinyl chloride, TYGON*, etc., may be used as test
 chambers or to^store effluents, but caution should be exercised in their use
 because they might introduce toxicants when new, or carry over toxicants
 from one test to another, if reused.  The use of large glass carboys  is
 discouraged for safety reasons.

 5.1.3  New plastic  products of a  type not previously used should be  tested
 for  toxicity before initial use by exposing the test organisms  in the test
 system where the material is  used.   Equipment  (pumps,  valves,  etc.)  which
 cannot be  discarded after each use  because of  cost,  must  be decontaminated
 according  to the cleaning procedures listed below  (Section  5.3  2)
 Fiberglass,  in  addition to the previously mentioned  materials,  can be used
 for  holding,  acclimating,  and dilution  water storage  tanks,  and in the water
 delivery system.  All  material should be  flushed or  rinsed  thoroughly with
 the  test media  before  using in the test.   Copper, galvanized material,
 rubber, brass,  and  lead must  not  come in  contact with culturing, holding
 acclimation,  or  dilution water, or with effluent samples  and test
 solutions.   Some materials, such  as  several types of neoprene rubber
 (commonly used for stoppers)  may  be toxic and  should be tested  before  use.

5.1.4  Silicone adhesive used to construct glass test chambers  absorbs some
organochlorine and organophosphorus pesticides, which are difficult to
remove.  Therefore,  as  little of the adhesive as possible should be in
contact with water.   Extra beads of adhesive inside the containers should be
i "mv V wU *                                                  , -:;
^Adapted from:  Peltier and Weber (1985).

-------
5.1.5  A good quality deionized water must be available in the laboratory.

5.2  TEST CHAMBERS

5 2 1  Test chamber size and shape are varied according to size of the test
organism.  Requirements are specified in each test.

5.3  CLEANING

531  New plasticware used for sample collection or organism exposure
vessels generally does not require rigorous cleaning, and it is usually
sufficient to soak the new containers overnight in seawater before use.   New
glassware also should be soaked overnight  in sea water.

5 3  2   It is recommended that  all sample containers, test vessels, pumps,
tanks,  and other equipment that has  come  in contact with effluent be washed
to remove surface contaminants after use,  as described  below.

     1.  Soak  15 minutes  in tap  water  and scrub with detergent, or  clean  in an
        automatic  dishwasher.
     2.  Rinse twice with  tap water.                     ...
     3   Carefully  rinse  once with  fresh  dilute  (20% V:V) nitric  acid or
      *  hydrochloric  acid to remove  scale,  metals  and  bases.   To prepare a
        20%  solution  of  acid,  slowly add 20 ml  of  concentrated  acid  to  80 ml
        of distilled  water.
     4.  Rinse  twice with deionized water.
     5.  Rinse  once with  full-strength acetone to remove organic  compounds.
     6*.  Rinse  well with  deionized  water.

 5.3.3  All  test chambers and  equipment must be thoroughly rinsed with  the
 dilution water immediately prior  to use in each test.

-------
                                   SECTION 6

                                 TEST  ORGANISMS
6.1  SPECIES
6.1.1 The organisms  used  in  the  short-term  tests  described  in  this  manual
are the sheepshead minnow, Cyprinodon  variegatus;  the  inland  silverside,
Menidia beryllina; the mysid, Mysidopsis  bahla^the  sea  urchin,  Arbacla
punctulata; and the  red macroalga, CJiampja  parvula*

6.2  SOURCE

6.2.1  These  test organisms  can  be cultured in the laboratory.   Culturing
and handling  procedures for  each organism are described  in  the respective
test method sections.

6.2.2.  Starter cultures  of  Champia £ar_v_u_l_a are available from the  U. S»
Environmental Protection  Agency, Environmental Research  Laboratory  -
Narragansett, South  Ferry Road,  Narragansett, RI 02882.

6.2.3  Sheepshead minnows, rnysids, and sea  urchins may be purchased from
commercial sources.  Inland  silversides may have to  be collected  in the
field.

6.2.4  If because of their source there is  any uncertainty  concerning the
identity of the organisms, it is advisable  to have them  examined  by a
taxonomic specialist to confirm  their  identification.  For  detailed guidance
on identification, see the individual toxicity test  methods.

6.3  SHIPMENT

6.3.1  Many states have strict regulations  regarding the importation of
non-native fishes.  Required clearances should be obtained  from  state
fisheries agencies before arrangements are  made for  the  interstate shipment
of sheepshead minnows or  silversides.

6.4  DISPOSAL

6.4.1  Because of possible toxicant or pathogen contaminations all test
organisms (including controls) must be humanely destroyed and disposed in an
appropriate manner.
                                      18

-------
                                   SECTION 7

                                DILUTION  WATER

7.1  Dilution water may be natural  seawater, hypersaline brine prepared from
natural seawater, or artificial seawater, depending on the test selected and
the objectives of the test:   {1}  if the objective of the test is to estimate
the inherent toxicity of the  effluent, a  dilution water of appropriate
salinity, prepared from deionized water and hypersaline brine or artificial
sea salts, is used; (2) if the objective  of the test is to estimate the
toxicity of the effluent in uncontaminated saline receiving water, the test
may be conducted using saline surface water collected from an uncontaminated
site or a saline water prepared with deionized water and hypersaline brine
or artificial sea salts; and  (3)  if the objective of the test is to
determine the additive effects of the discharge on already contaminated
receiving water, the test is performed using saline surface water collected
as close as possible to the outfall, but  outside the zone contaminated by
the outfall.

7.2  The selection of dilution water may  limit the maximum concentration of
effluent that can be used in the test.

7.3  When off-site, uncontaminated, receiving water is used, it should be
collected immediately prior to the test-,  but never more than 96 h before the
test begins.  Except where the water is used within 24 h, it should be
chilled to 4°C during or immediately following collection, and maintained
at that temperature until  warmed again before use.

7.4  Where toxicity-free dilution water is required in a test, the water is
considered adequate if test acceptability criteria (survival, growth,  and
reproduction) are met.

7.5  Artificial  seawater is to be used only if specified in the method.
EMSL-Cincinnati  has found  FORTY FATHOMS^ artifical sea salts (Marine
Enterprises,  Inc.,  8755 Mylander Lane, Baltimore,  Maryland 21204;  phone:
301-321-1189} suitable for maintaining and spawning sheepshead minnows, and
for their use in the larval  survival and growth test,  and for maintaining,
spawning,  and testing sea  urchins.  The USEPA Region 6 Houston Laboratory
has successfully used HW MARINEMIX1* (Hawaiian Marine Imports Inc.,  P.O.
Box 218687,  Houston,  Texas 77218,  phone 713-492-7864)  sea salts to maintain
and spawn sheepshead minnows,  and perform the embryo-larval  survival  and
teratogenicity test.
                                      19

-------
                                 SECTION 8
            EFFLUENT AND RECEIVING WATER SAMPLING AND SAMPLE HANDLING
  8.1  EFFLUENT SAMPLING
                                 1979C? "cond^™ '! that "»C'"«' '»

        -
              S-SS*
 8.1.3  Aeration during collection and transfer  of effluents «hmii/t
 nunimized to reduce the loss of volatile chemicals            ?d
 8.2  RECEIVING WATER SAMPLING        " — •
8.3  SAMPLE HANDLING AND PRESERVATION

                                                           "PDEs

-------
 after removal  from the sampling device.   Composite samples should be chilled
 during collection,  where possible.   Samples must be chilled after collection
 and maintained at 4°C until  warmed  up for use.

 8.3.2 Samples  Used  in On-Site Tests                                   '

 8.3.2.1   Samples  collected for on-site tests should be used within 24 h.

 8.3.3  Samples used in Off-Site Facilities

 8.3.3.1   Samples  collected.for off-site  toxicity testing  are to  be chilled
 to  4°C when  collected,  shipped iced to the central  laboratory, and there
 transferred  to a  refrigerator (4°C)  until  used.   If the persistence of the
 toxicity  of  the sample  is  not known,  the test should begin within 36 h of
 sample removal  from the  sampling  site.   Every effort must  be made to
 initiate  the test with  an  effluent  sample  on the day of arrival  in the
 laboratory.

 8.3.3.2   Samples  may be  shipped  in  4-L  tl-gal) glass jugs,  CUBITAINER$R,
 or  new plastic  "milk"  jugs.   All  sample  containers  should  be rinsed with
 source water before  being  filled  with  sample.  Glass jugs  can be  cleaned and
 reused (see  p.  17),  whereas  CUBITAINERS^ and plastic jugs  are not reused.
 Plastic containers  used  for  effluents  or toxic surface  water samples should
 be  punctured after  use to  prevent reuse.

 8.4   SAMPLE  PREPARATION

 8.4.1  Adjust  the sample salinity to the  level appropriate  to objectives of
 the study using hypersaline  brine or artificial  sea  salts.

 8.4.2  If necessary, effluent  and surface waters may be filtered  through a
 30 urn  plankton  net to remove  indigenous organisms that may  attack  or  be
 confused with the test organisms.   It may be necessary to first
 coarse-filter the dilution and/or waste water through a NYLONR sieve
 having 2- to 4-mm holes to remove debris and/or break up large floating or
 suspended solids.   Since filtering may increase the  dissolved oxygen  (DO) in
 an effluent, the DO should be determined prior to filtering.  Low  dissolved
 oxygen concentrations will indicate a potential  problem in  performing  the
 test.

8.4.3  The dissolved oxygen concentration (DO) in the dilution water  should
be near saturation prior to use.  Aeration will  bring the DO and other gases
 into equilibrium with air, minimize oxygen demand, and stabilize the  pH.

8.4.4  If the dilution water and effluent must be warmed to bring them to
the prescribed test temperature, supersaturation of the dissolved gases may
become a problem.   To prevent this problem, the effluent and dilution water
are heated to 25°C and checked for dissolved oxygen  (DO) with a probe.  If
the DO exceeds  100% saturation based on temperature and salinity, the
solutions are aerated vigorously with an air stone (usually 1-2 min)  until
the DO is lowered  to 100% saturation (Table 1).
                                      21

-------
Table 1.   OXYGEN SOLUBILITY (MG/L) IN WATER AT EQUILIBRIUM WITH AIR AT
          760 MM HG (AFTER RICHARDS AND CQRWIN,  1956)
TEMP
0
1
2
3
4
5
6
8
10
12
14
16
18
20
22
24
26
28
30
32
SALINITY (o/oo)
0
14.2
13.8'
13.4
13.1
12.7
12.4
12.1
11.5
10.9
10.5
10.0
• 9.6
9.2
8.9
8.6
8.3
8.1
7.8
7.6
7.3
5
13
13
13,
12.
12.
12.
11.
11.
10.
10.
9.
9.
9.
8.
8.
8.
7.
7.
7.
7.
10
.8
.4
.0
.7
,3
,0
.7
,2
7
2
7
3'
0
6
4
1
8
6
4
1
13
13
12.
12,
12.
11.
11.
10.
10.
9.
9.
9.
8.
8.
8.
7.
7.
7.
7.
6.
.*
.0
.6
.3
.0
.7
.4
,8
3
9
5
1
7
4
1
8
6
4
1
9
15
12.9
12.6
12.2
11.9
11.6
11.3
11.0
10.5
10.0
9.6
9.2
8.8
8.5
8.]
7.9
7.6
7.4
7.2
6.9
6.7
20 25
^2.5 12.'
12.2 11.8
M .9 11.5
:i.e 11.2
;<.3 10.9
M.O 10.6
10.7 10.3
10.2 9.8
9.7 9.4
9.3 9.0
8.9 8. fa ,
8.5 8.3
8.2 8,0
7.9 7.7
7.6 7.4
7.4 7.2
7.2 7.0
7.0 6.8
6.7 6.5
6.5 .6,3
30
IT
11
11
10.
.7
.4
.1
.8
10.5
10.2
10.
9.
9.
8.
8.
8.
7.
7.
7.
6.
6.
6.
6.
6.
,0
i;
h -j
1
7
3
0
7
4
2
9
7
5
3
1
35 4Q
11.2 ' TO. 3
11.0 TO. 6
10.7 ^0.3
10.4 ]G. 0
10.1 9.3
9.8 • 9.5
9.6 9.3
9.2 8,9
8.8 8.5
8.4 8.1
8.1 7.8
7.7 7,5
7.b 7.2
7.2 6.9
6.9 6.7
5.7 6.5
6.5 6.3
6.3 6.1
6.1 5.9
5.9 5.7
43
1C. 6
i0.3
10.0
9\8
9.5
9,3
9.1
8.7
8.3
7.9
7.6
7.3
T •
6.8
6.6
6,4 -
6.1
6.0
5.8
5.6
                                      22

-------
                                    SECTION 9

                CHRONIC  TOXICITY  TEST  END  POINTS  AND  DATA ANALYSIS

 9.1  END  POINTS

 9.1.1  The objective  of chronic  aquatic toxicity tests with  effluents and pure
 compounds is to estimate the  highest  "safe" or "no-effect  concentration" of
 these  substances.  For  practical reasons,  the parameters observed  in these
 tests  are usually  limited  to  hatchability, survival, growth,  and reproduction,
 and the results of the  tests  are usually  expressed in terms  of  the highest
 toxicant  concentration  that has  no  statistically significant  observed effect
 on these  parameters.  The  terms  currently used to define the  end points
 employed  in the rapid,  chronic and  sub-chronic toxicity tests have been
 derived from the terms  previously used for full  life-cycle tests.  As shorter
 chronic tests were developed, it became common practice to apply the same
 terminology to  the end  points.   The primary terms in current  use are as
 follows:

 9.1.1.1  Safe Concentration - The highest  concentration of toxicant that will
 permit normal propagation  of fish and other aquatic  life in receiving waters.
 The concept of  a "safe  concentration" is  a biological concept,  whereas the
 "no-observed-effect concentration"  (below) is a  statistically defined
 concentration.

 9.1.1.2  No-Observed-Effect Concentration  (NOEC) - The highest  concentration
 of toxicant to  which organisms are exposed in a  full life-cycle or partial
 life-cycle test, that causes no observable adverse effects on the test
 organisms - i.e., the highest concentration of toxicant in which the values
 for the observed parameters are not statistically significantly different from
 the controls.   In the discharge permit program the "safe concentration" is
 currently defined as the "no-observed-effect concentration."

 9.1.1.3.  Lowest-Observed-Effect Concentration (LOEC) - The lowest
 concentration of toxicant to which organisms are exposed in a life-cycle or
 partial life-cycle test, which causes adverse effects on the test organisms -
 i.e.  where the  values for the observed parameters are statistically
 significantly different from the controls.

9.1.1.4.  Maximum Acceptable Toxicant Concentration (MATC)  - An undetermined
concentration within the interval bounded by the NOEC and LOEC that is
presumed safe by virtue of the fact that no adverse effect was observed.

9.1.1.5. - Chronic Value  (ChV)  - A point estimate of the presumably safe
 (no-effect)  concentration, lying between the NOEC and LOEC, and derived by
calculating  the geometric mean of the NOEC and LOEC.   The geometric .mean (chV)
 is assumed to be the "safe" concentration.

-------
                                                               I" I   ;- •  {
t  . t .-   r
                                        I I •'.
                                                                    V ' f
                                                   (  - i   ' C  f-
                                                                                      . t  (I .' -j
                                           r;i:  I
                                                               d"6"'  rrnr  <\   r i
 Hi-.'.  ,
:  ".rrs
                                                                               T;--.r
                                                             >c  i  7C
                                                                    H
                                                                             ••.  -i.-j  ,-,'   - ', r,    j'-r    " •
                                                                          f.    !.:.(»  -3';V  rrO   >• f f £
                             — V  '.*
                          f     /

                          i ..  i t,:i''.

-------
 9.4 REPLICATION AND SENSITIVITY
                               an
                              th.
            «•=; SfS'sHH > •

                        sesa
9.4.3 CHOICE OF ANALYSIS AND MULTIPLE NOEC'S
               26

-------
   9.5   ANALYSIS  OF GROWTH  AND  REPRODUCTION  DATA

   9.5.1   Growth  data from  the  sheepshead minnow  and  inland  silverside  larval
   survival and growth tests, and the mysid  survival, growth, and fecundity test
   are analyzed using hypothesis testing according to the flow chart  in
   Figure  2.   (Note  that the nonparametric tests  can be used only if  at least
   four replicates were used at each toxicant concentration).

   9.5.2  Fecundity data from the mysid tests may be analyzed either by
   hypothesis testing (after an arc sine transformation) according to the flow
   chart in Figure 2, or by generating a point estimate.  The point estimate may
  be obtained by using Profait Analysis (Finney, 1971),  if appropriate (see
  discussion below).  An adjustment should  be added to  the Probit Analysis for
  the percentage of females without eggs in  the controls.

  9.5.3   Reproduction  data  from the Champia  Test  are  analyzed  using  hypothesis
  testing  as  illustrated  in Figure  2.                                  K

  9.6 ANALYSIS  OF  SEA URCHIN FERTILIZATION  DATA

  9.6.1  Data  from  the sea  urchin fertilization test may be  analyzed  by
  hypothesis testing after  an arc sine  transformation according to the flow
  chart  in Figure 2.  The fertilization data from the sea urchin test may also
  be  analyzed  by generating a point estimate with Probit Analysis, after an
  adjustment for the infertility rate in the controls/  if Probit Analysis is
  appropriate  (see discussion below).

  9.7  ANALYSIS OF MORTALITY DATA

  9.7.1   Mortality data from the sheepshead  minnow and inland silverside larval
  survival  and growth tests, the sheepshead  minnow embryo-larval  survival  and
 teratogemcity test,  and the mysid survival,  growth,  and reproduction test are
 analyzed  by Probit Analysis,  if appropriate (see discussion below).   The
 mortality data  can also  be analyzed by hypothesis  testing,  after an  arc  sine
 transformation  (see Appendix), according to the  flow chart  in Figure 2.

 9.8  DUNNETT'S  PROCEDURE

 9.8.1   Dunnett's Procedure consists of an analysis of  variance (ANOVA) to
 determine the error term,  which is  then used  in  a multiple  comparison method
 for  comparing each  of the  treatment means with the control  mean, in  a series
 of paired tests.   Use of Dunnett's  Procedure requires  at least two replicates
 per  treatment and  an equal number of data points (replicates) for each
 concentration.   In  cases where the numbers of data points for each
 concentration are not equal, a t-test may be performed with Bonferroni's
 adjustment for multiple comparisons, instead of using Dunnett's Procedure (see
 Appendix),

 9.8.2  The assumptions upon which the use of Dunnett's  Procedure is contingent
are that the observations within treatments  are independent and  normally
distributed,  with homogeneity of variance.   Before analyzing the  data,  the
assumptions  are  checked  using the procedures provided in  the Appendix.

                                       27

-------
   t
that this
                     1'^  °f-t-8 Se"?"iv1t^ of tie analysis should  be provided
                     ." snj'srr4ss",s; ss s
                    difference represents  for  a given test.
  9.8.4  The estimate of the safe concentration derived from this test- i

                                                               "
                                                              d 2"
                                                -ncen   t      i 6th      ers
  9.9   BONFERRONI'S T-TEST
  Thus, Dunnett's  Procedure is a more powerful test

  9.10  STEEL'S  MANY-ONE RANK TEST

  9.10.1  Steel's  Many-One Rank Test  is a multiple comparison method for
  comparing  several treatments with a  control which is  similar to Dunnett's
  nL°Led>e>  eT,Cep5 5hat n 1s not "ec"sary to meet  the a  umption ?or
  normality.  The  data are ranked, and the analysis is  performed on the rank,
 rather than on the data themselves.  If the data are  normally o? near Iv
 normally distributed, Dunnett's  Procedure would  be  more sensitive (would
 detect smaller differences  between the treatments and contro )    For  d ta that
 are not  normally distributed, Steel's Many-One Rank Test can be mu2h  more
 eff cient  (Hodges and Lehmann, 1956).  It is  necessary to have  at leart four
 replicates per toxicant concentration to use  Steel's test.   The sensltfvS of
                                                           htTV1
9.11  WILCOXON  RANK SUM TEST

9.11 1  The Wilcoxon Rank Sum Test  is a nonparametric test for comoarina a
treatment with  a control.  The data are ranked and the analysis proS
exactly as in Steel's Test except that Bonferroni's adjustment fSr muHiole
SnrtHS°nS 1S  US6d  nstead °f Steel's tables"  "*™ Steel's test can be sed
(when there are equa  numbers of data points per toxicant concentration)  it
w 11 be more powerful  (able to detect smaller differences as statist  caiv
signif1Cant) than the  Wilcoxon Rank Sum with Bonferroni's adjustment
                                     28

-------
  the Appendix.


  9.12  PROBIT ANALYSIS
                                                       « «» NOEC.  A
                                      i icoxon Rank  Sum Test 1S  provided in
                                                       fro,
concentraton oT^canT^ec   9

ProV1de  a confidence  interval for  the Istimatl   Prnhit A   i6St Or9an1sms
normal distribution of loq tolerant !nH ?!;!'  ?roblt ^nal>S7S assumes a
responses.  To use Prob ttAnffil   « lea  ^^ ?f the 1nd^1dual
obtained.                «naiyS1S, at least two partial mortalities must be
        r
ext™« caution.


                              Mtnatrt,
                                             results
                                                              «s«l ,ltt
sssr
                                                             „
                                                      s a: sss1



-------
               DATA  (SURVIVAL.  GROWTH.  REPRODUCTION.  ETC.)
                                 I
                           TRANSFORMATION?
  ENDPOINT ESTIMATE
  EC1. EC5.EC10. EC50
SHAPIRO-WILKS TEST
             NORMAL DISTRIBUTION
HOMOGENEOUS  VARIANCE
        NO
       I
                     NON-NORMAL DISTRIBUTION
                           BARTLETT'S TEST
                                              HETEROGENEOUS
                                                VARIANCE
                         NO STATISTICAL ANALYSIS
                               RECOMMENDED
                                                  NO
                              4 OR MORE
                             REPLICATES?
                                                   YES
               EQUAL NUMBER OF
                 REPLICATES?
            EQUAL NUMBER OF
              REPLICATES?
YES
                                     i
               YES
T-TEST KITH
BONFERRONI
ADJUSTMENT



DUNNETT'S
TEST

STEEL'S MANY-ONE
RANK TEST



HILCQXON
TEST
BONFERRONI

RANK SUM
WITH
ADJUSTMENT

                          ENDPOINT ESTIMATES
                               NOEC. LOEC
Figure 2. Flow chart for  statistical  analysis of test data

                             JO

-------
                                    SECTION 10

                               REPORT PREPARATION!

   Th0  JJ*to?icit> dat? *re reported,  together with other appropriate data,
   The  following  general format and  content are recommended for the report:

   10.1   INTRODUCTION
         1,
        2.
        3,
        4,
        5.
Permit  number
Toxicity testing requirements of permit
Plant location                               ^v:K;s
Name of receiving water body
Contract Laboratory (if the test was performed under contract)
a. Name of firm
b. Phone number
c. Address
  10.2  PLANT OPERATIONS

        1.  Product(s)
        2.  Raw materials
        3.  Operating schedule
        4.  Description of waste treatment
        5.  Schematic of  waste treatment
        6.  Retention time (if applicable)
        7.  Volume of waste flow (MGD, CFS, GPM)
 10.3  SOURCE OF EFFLUENT AND DILUTION WATER

       1. Effluent Samples
          a. Sampling point
          b. Collection dates and times
          c. Sample collection method
          d. Physical and chemical  data
          Surface  Water  Samples
          a.  Sampling  point
          b.  Collection  dates  and  times
          c.  Sample  collection method
          d.  Physical  and chemical data
          e.  Tide  stages
'Adapted from: Peltier and Weber (1985).  Prepared by Lee Tebo and William
Peltier  Environmenta  Services Division, U.S. Environmental Protection
Agency, Athens, Georgia.              ,

-------
              Dilution  Water Samples
              a.  Source
              b.  Collection  date  and time
              c.  Pretreatment
              d.  Physical and chemical characteristics
   10.4  TEST METHODS
  1.
  2.
  3.
  4.
  5.
  6.
  7.
  8.
  9.
10.

11.
12.
             Toxicity test method used
             End^point(s)  of test
                                         method, if

             Date  and  time  test  terminated
             Type  of test chambers
             Volume of solution  used/chamber
            Number of organisms/test chamber
            Numnnr> rt-P«««ij__J._ .    .
                                     chambers/treatment
                                     i ^ **in^ / i —„_.__   .
            Test temperature (mean and range)
            Specify !f aeration was needed
                                                       and salinity mean
  10.5   TEST  ORGANISMS
        1
        2,
        3,
        4,
        5.
        6.
        7.
   Scientific  name
   Age
   Life stage
                           (where
   Diseases and treatment  (where applicable)
   Taxonomic key used for species identification
 10.6   QUALITY  ASSURANCE.
        1,
        2.
        3,
       4,
   Standard  toxicant  used,  and source
   Date and  time  of most  recent  test
   uilution  water used  in test
   Results (LC50  or,  where  applicable,  NOEC  and/or
   Phsical  a             --.    ''    LU  ana/or
       c  ML.  • -- *•	•« "i » wiicr'e ctppncao 6
       5. Physical and chemical methods used
10.7  RESULTS

      si
                                                           s
         Pr.v,d« ,.M, of
                                    32

-------
                                     SECTION 11

                                  TEST METHOD 1.2

                     SHEEPSHEAD MINNOW (CYPRINODON VARIEGATUS)
                          LARVAL SURVIVAL AND GROWTH TEST
                                    METHOD ]004

  1,  SCOPE AND APPLICATION

  1.1  This method estimates the chronic toxicity of effluents and receiving
 waters to the sheepshead minnow (Cyprinodon variegatus), using newly hatched
  larvae in a seven-day, static-renewal test.  The effects include the
 synergistic, antagonistic, and additive effects of all the chemical, physical,
 and biological components which adversely affect the physiological and
 biochemical functions of the test species.

 1.2  Detection limits of the toxicity of an effluent or pure substance are
 organism dependent.

 1.3  Single or multiple excursions in toxicity may not be detected using 24-h
 composite samples.   Also, because of the long sample collection period
 involved  in composite sampling, and because the test chambers are not sealed,
 highly volatile and highly degradable toxicants in the source may not be
 detected  in the test.

 1.4  This method should be restricted to use by,  or under the supervision of,
 professionals  experienced in  aquatic toxicity testing.

 2.   SUMMARY  OF METHOD

 2.1  Larvae  (preferrably  less than  24 h  old)  are  exposed  in  a static  renewal
 system  for  seven  days to  different  concentrations  of  effluent or  to receiving
 water.  Test results  are  based  on the survival  and  weight of  the  larvae  in
 test solutions, compared  to controls.

 3.  DEFINITIONS

    (Reserved for  addition of  terms at  a  later date).

 4.  INTERFERENCES

 4.1  Toxic substances may be  introduced by contaminants, in dilution water,
 glassware, sample hardware, and testing equipment (see Section 5, Facilities
 and Equipment).
     format used for this method was taken from Kopp, 1983.
2This method was adapted in part from Horning and Weber, 1985, and was based
on the method of Hughes, Heber, Schimmel, and Berry, 1987, Environmental
Research Laboratory, USEPA, Narragansett, Rhode Island.

                                       33

-------
  4.2  Adverse effects of low dissolved oxygen concentrations (DO), hiah
  concentrations of suspended and/or dissolved solids, and extremes of pH, may
  mask the effects of toxic substances,                                H     y

  4.3  Improper effluent sampling and handling may adversely affect test
  results  see Section 8, Effluent and Receiving Water Sampling  and Sample
  Hand I ing ) .                                                            "

  4.4 Pathogenic  and/or predatory organisms  in  the  dilution  water  and
  effluent may affect  test organism survival,  and  confound  test  results.

  4.5 Food added  during  the  test  may  sequester  metals and  other toxic
  substances and reduce  the apparent toxicity  of the test substance.  However
  in  a growth  test the nutritional  needs of the  organisms must be satisfied!
  even if  feeding  has the potential to confound  test results.     bdL7STiea>

 .5.  SAFETY

 5.1  See Section 3, Health and Safety.

 6.  APPARATUS AND EQUIPMENT

 6.1 Facilities for  holding  and acclimating  test organisms.

 6.2  Brine shrimp culture unit — see 7.14  below.

 6.3^ Sheepshead minnow  culture unit  —  see  Paragraph  7.15  below    The
 maximum  number of larvae required per test will range from a maximum of  360,
 I   I  -f in6^  USed ln each  Of  four replicates, to a minimum of  180 per
 I    L •  ™ larvae  are  used  in  each of  three  replicates.   It is preferable
 to  obtain the  test  organisms from an  inhouse  culture unit.   If it  is not
 feasible  to culture fish inhouse,  embryos or  newly hatched larvae  can be
 obtained  from other sources  if  shipped  in well oxygenated  saline water in
 insulated containers.

6.4  Samplers  - automatic sampler, preferably with sample coolinq
capability,  that can collect a 24-h composite sample of 5 L.


(25+
                            °r  e(*u1valent  ^cility with  temperature  control
6.6_ Water purification system - MTIlipore Super-Q, deionized water  (DIJ or
equi va lent .

6.7  Balance — Analytical, capable of accurately weighing to 0.0001 g.
   nh    hiH                S " f°r Check1"9 Performance of balance.
Weights should bracket the expected weights of the weighing boats and the
expected weights of the weighing boats plus fish.
                                      34

-------
6.9  Drying oven — 1050C, for drying larvae.
6.10  Air pump — for oil-free air supply.
6.11  Air lines, and air stones — for aerating water containing embryos or
larvae, or for supplying air to test solutions with low DO.
6.12  pH and DO meters — for routine physical and chemical measurements.
Unless the test is being conducted to specifically measure the effect of one
of the above parameters, a portable, field-grade instrument is acceptable.
6.13  Standard or micro-Winkler apparatus — for determining DO (optional).
6.14  Dissecting microscope — for checking embryo viability.
6.15  Desiccator — for holding dried larvae.
6.16  Light box -- for counting and observing larvae.
6.17  Refractometer -- for determining salinity.
6.18  Thermometers, glass or electronic, laboratory grade -- for measuring
water temperatures.
6.19  Thermometers, bulb-thermograph or electronic-chart type « for
continuously recording temperature.
6.20  Thermometer, National Bureau of Standards Certified (see USEPA METHOD
170.1, USEPA, 1979) -- to calibrate laboratory thermometers.
6.21  Test chambers —  four (minimum of three) for each concentration and
control.  To avoid potential contamination from the air and evaporation of
water from the test solutions, the chambers should be covered during the
test.
6.22  Beakers — six Class A, borosilicate glass or non-toxic plasticware,
1000 ml for making test solutions.
6.23  Wash bottles — for deionized water, for washing embryos from
substrates and containers, and for rinsing small glassware and instrument
electrodes and probes.
6.24  Crystallization dishes, beakers, culture dishes, or equivalent — for
incubating embryos.
6.25  Volumetric flasks and graduated cylinders — Class A, borosilicate
glass or non-toxic plastic labware, 10-1000 ml for making test solutions.
6.26  Separatory funnels, 2-L — two to four for culturing Artemia.
6.27  Pipets, volumetric — Class A, 1-100 ml.

-------
 6.28  Pipets, automatic ™ adjustable,  1-100 ml.
 6.29  Pipets, serological — 1-10 ml, graduated.
 6.30  Pipet bulbs and fillers — PROPIPETR, or equivalent.
 6.31  Droppers, and glass tubing with fire polished edges, 4 mm ID — for
 transferring larvae.
 6.32  Siphon with bulb and clamp — for cleaning test chambers.
 6.33  Forceps — for transferring dead larvae to weighing boats.
 6.34  NITEXR or stainless steel  mesh sieves,  < 150 urn, 500 urn, 3-5 mm —
 for collecting Artemia nauplii  and fish embryos, and for spawning baskets,
 respectively.   (Available from  Sterling Marine Products, 18 Label  Street,
 Montclair,  NJ  07042; phone 201-783-9800).
 7.   REAGENTS AND CONSUMABLE  MATERIALS              -,.-.-,,.
 7.1   Sample containers — for sample shipment and storage {see Section 8.
 Effluent  and Receiving Water  Sampling and  Sample Handling).
 7.2   Data sheets (one  set  per test)  -- for data  recording.
 7.3   Vials,  marked  —  18-24 per  test,  containing 4%  formalin  or 70% ethanol,
 to preserve  larvae.  (Optional).
 7.4   Weighing  boats, aluminum —  18-24 per test.
 7.5   Tape,  colored  —  for  labelling  test chambers.
 7.6   Markers, water-proof  —  for marking containers, etc.
 7.7   Buffer, pH  7,  (or as  per instructions  of  instrument manufacturer) —
 for  standards and calibration check  (see USEPA Method  150.1,  USEPA,  1979).
 7.8  Membranes and filling solutions for dissolved oxygen probe  (see USEPA
 Method 360.1, USEPA, 1979), or reagents —  for modified  Winkler  analysis.
 7.9  Laboratory quality control  samples and standards  — for  calibration of
 the above methods.
 7.10  Reference toxicant solutions (see Section 4, Quality Assurance).
 7.11  Formalin (4%)  or 70% ethanol — for use as a preservative for the fish
 larvae.
 7.12  Reagent water— defined as distilled or deionized water that does not
contain substances which are toxic to the-test organisms  (see paragraph 6.6
above).

-------
7.13  Effluent, surface water, and dilution water -- see Section 7, Dilution
Water, and Section 8, Effluent and Surface Water Sampling and Sample
Handling.

7.13.1  Saline test and dilution water — The salinity of the test water
must be in the range of 20 to 32 °/oo.  The salinity should vary by no
more than ± Z o/oo among the chambers on a given day.  If effluent and
receiving water tests are conducted concurrently, the salinities of these
tests should be similar.  This test is not recommended for salinities less
than 20 o/oo.

7.13.2  The overwhelming majority of industrial and sewage treatment
effluents entering marine and estuarine systems contain little or no
measurable salts.  Exposure of sheepshead minnow larvae to these effluents
will require adjustments in the salinity of the test solutions.  It is
important to maintain a constant salinity across all treatments.  In
addition, it may be desirable to match the test salinity with that of the
receiving water.  Two methods are available to adjust salinities -- a
hypersaline brine derived from natural seawater or artificial sea salts.

7.13.3  Hypersaline brine (100 °/oo salinity):  Hypersaline brine (HSB)
has several advantages that make it desirable for use in toxicity testing.
It can be made from any high quality, filtered seawater by evaporation, and
can be added to the effluent or to deionized water to increase the
salinity.  HSB derived from natural seawater contains the necessary trace
metals, biogenic colloids, and some of the microbial components necessary
for adequate growth, survival, and/or reproduction of marine and estuarine
organisms, and may be stored for prolonged periods without any apparent
degradation.  However, if 100 °/0o salinity HSB is as a diluent, the
maximum concentration of effluent that can be tested will be 80% at
20 o/oo salinity and 70% at 30 o/oo salinity.

7.13.3.1  The ideal container for making brine from natural seawater is one
that (1) has a high surface to volume ratio, (2) is made of a non-corrosive
material, and (3) is easily cleaned (fiberglass containers are ideal).
Special care should be used to prevent any toxic materials from coming in
contact with the seawater being used to generate the brine.  If a heater is
immersed directly into the seawater, ensure that the heater materials do not
corrode or leach any substances that would contaminate the brine.  One
successful method used is a thermostatically controlled heat exchanger made
from fiberglass.  If aeration is used, use only oil-free air compressors to
prevent contamination.

7.13.3.2  Before adding seawater to the brine generator, thoroughly clean
the generator, aeration supply tube, heater, and any other materials that
will be in direct contact with the brine.  A good quality biodegradable
detergent should be used, followed by several (at least three) thorough
deionized water rinses.

7.13.3.3  High quality (and preferably high salinity) seawater should be
filtered to at least 10 urn before placing into the brine generator.  Water
should be collected on an incomming tide to minimize the possibility of
contamination.

-------
  7.13.3.4   The  temperature  of the  seawater is  increased  slowly to
  The water  should  be  aerated  to  prevent  temperature  stratification  and  to
  increase water evaporation.   The  brine  should  be  checked  daily (depending  on
  volume being generated)  to ensure that  the  salinity does  .not  exceed
  100 °/oo and that the  temperature does  not  exceed WC.   Additional
  seawater may be added  to the  brine to obtain the  volume of  brine required.

  7.13.3.5  After the  required  salinity is  attained,  the brine  should be
  filtered a second time through  a  1  urn filter and  poured directly into
  portable containers, such  as  20-L  (5 gal) cubitainers or  polycarbonate water
  cooler jugs.  The containers  should be  capped and labelled with  the date the
  brine was generated and  its salinity.   Containers of brine should be stored
  in the dark and maintained at room temperature until used.

  7.13.3.6  If a source of hypersaline brine is available, test solutions can
 be made by following the directions below.  Thoroughly mix together the
 deionized water and brine before adding  the effluent.

 7.13.3.7   Divide the salinity of the hypersaline brine by the expected  test
 salinity  to determine the proportion of  deionized  water  to brine*  For
 example,  if the salinity of the  brine is 100 o/oo  and the  test is to  be
 conducted  at  20 °/oo, 100 °/oo divided by 20 
-------
7.14   BRINE  SHRIMP  (ARTEMIA)  NAUPLII  (see  Peltier  and  Weber,  1985).

7.14.1   Newly-hatched Artemia nauplii  are  used  as  food for  sheepshead  minnow
larvae in  toxicity  tests  and  in  the maintenance of continuous  stock
cultures.  Although  there are many commercial sources  of  brine shrimp  cysts,
the Brazilian  or  Colombian strains are currently preferred  because the
supplies examined have had low concentrations of chemical residues and
produce  nauplii of  suitably small  size.   (One source that has  been found  to
be acceptable  is  Aquarium Products, 180L Penrod Ct., Glen Burnie, Maryland
21061, phone 800-368-2507).

7.14.2   Each new  batch  of Artemia  cysts must be evaluated for  size (Vanhaecke
and Sorgeloos,  1980,  and  Vanhaecke et  al.,  1980) and nutritional  suitability
(see Leger et  al.,  1985,  1986) against known suitable  reference cysts  by
performing a side by  side larval growth test using  the "new" and  "reference"
cysts.   The  "reference" cysts  used in  the  suitability  test  may be a
previously tested and  acceptable batch of  cysts, or may be  obtained from  the
Quality  Assurance Branch,  Environmental Monitoring  and Support Laboratory,
Cincinnati, Ohio.  A  sample of newly-hatched Artemia nauplii from each new
batch of cysts  should  be  chemically analyzed.   The Artemia  cysts should not
be used  if the concentration  of total  organic chlorine exceeds 0.15 ug/g  wet
weight or the  total concentration  of organochlorine pesticides plus PCBs
exceeds  0.30 ug/g wet weight.  (For analytical  methods see  USEPA, 1982).

7-14.3   Artemia nauplii are obtained as follows:

    1.   Add 1  L of seawater, or a  solution prepared by adding  35.0 g
         uniodized salt  (NaCl) or artificial sea  salts  to 1  L deionized water,
         to a 2-L  separatory funnel, or equivalent.
    2.   Add 10 ml Artemia  cysts to the separatory funnel and aerate for 24 h
         at room temperature.   (Hatching time varies with incubation
         temperature and the geographic strain of Artemia used).  (See Peltier
         and Weber, 1985,  and ASTM  designation E1203, 1987,  for details on
         Artemia culture and quality control).
    3.   After 24  h,  cut off the air supply in the separatory funnel.   Artemia
         nauplii are phototactic,  and will concentrate  at the bottom if a dark
         cloth or  paper towel is placed over the top of the  separatory funnel
         for 5-10 min.  To  prevent mortality, do not leave the  concentrated
         nauplii at the bottom of the funnel more than  10 min without  aeration.
    4.   Drain the nauplii  into a cup or funnel  fitted with a < 150 urn Nitex
        or stainless steel screen,  and rinse with seawater or  equivalent
        before used.

7.14.4  Testing Artemia nauplii as  food for toxicity test organisms.

7.14.4.1   The primary criterion for acceptability of each new  supply  of brine
shrimp cysts is the  ability of the  nauplii  to support good survival  and
growth of the sheepshead minnow larvae (see Paragraph  12., ACCEPTABILITY OF
TEST RESULTS).   The  larvae used to  evaluate the  suitability of the brine
shrimp nauplii  must  be of the same  geographical  origin, species,  and  stage
                                      39

-------
of development as those used routinely in the toxicity tests.  Sufficient
data to detect differences in survival and growth should be obtained by
using three replicate test vessels, each containing a minimum of 15 larvae,
for each type of food.

7 14 4 2  The feeding rate and frequency, test vessels, volume of control
water, duration of the test, and age of the nauplii.at the start of the
test, should be the same as used for the routine toxicity tests.

7.14.4.3  Results of the brine shrimp nutrition assay, where there are only
two treatments, can be evaluated statistically by use of a t-test.  The
"new" food is acceptable if there are no statistically significant
differences in the survival and growth of the larvae fed the two sources of
nauplii.

7.15   SHEEPSHEAD MINNOWS

7.15.1  Brood Stock

7.15.1.1  Adult sheepshead minnows for use as brood stock may be obtained by
seine in Gulf of Mexico and Atlantic coast estuaries, from commercial
sources, or from young fish raised to maturity in the laboratory.  Feral
brood stocks and first generation laboratory fish are preferred, to minimize
inbreeding.

7 15.1.2  To detect disease and to allow time for acute mortality due to the
stress of capture, field-caught adults are observed in the laboratory a
minimum of two weeks before using as a source of gametes.  Injured or
diseased fish are discarded.

7 15.1.3  Sheepshead minnows can be continuously cultured in the laboratory
from eggs to adults.  The larvae, juvenile, and adult fish should be kept in
appropriate size rearing tanks, maintained at ambient laboratory
temperature.  The larvae should be fed sufficient newly-hatched Artemia.
nauplii daily to assure that live nauplii are always present.  Juveniles are
fed frozen adult brine shrimp and a commercial flake food, such as TETRA
SM-8QR, available from Tetra Sales (U.S.A), 201 Tabor Rd, Morris Plains,
New Jersey 07950, phone 800-526-0650, or MARDEL AQUARIANR Tropical Fish
Flakes, available from Mardel Laboratories, Inc., 1958 Brandon Court,
Glendale Heights, Illinois 60139, phone 312-351-0606, or equivalent.  Adult
fish (age one month) are fed flake food three or four times daily,
supplemented with frozen adult brine shrimp.

7 15 1 3 1  Sheepshead minnows reach sexual maturity in three-to-five months
after hatch, and have an average standard length of approximately 27 -mm for
females and 34 mm for males.  At this time, the males begin to exhibit
sexual dimorphism and initiate territorial behavior.  When the fish reach
sexual maturity and are to be used for natural spawning, the temperature
should be controlled at 18-200C.

-------
                                              zsssys
   7.15.1.5  The  system is equipped with an undergravel or outside bioloaical
   K wMtir- tsts ?nr^r/r»jsjj's;;er'
                                            , r,,a
  7.15.2  Obtaining Embryos for Toxicity Tests 1
  7.15.2.1  Embryos  can be shipped  to the laboratory from an outside source or
  obtained from adults held in the  laboratory.   Ripe eggs can be obtained
  e  ther by natural  spawning or by  intraperitoneal  injection of the females
  with  uman chorionic gonadotrophin (HCG) hormone, available froTuntted
  States Biochemical Corporation, Cleveland,  Ohio 44128, phone 216-765-5000
  If the culturing system for  adults is temperature controlled, natural
  spawning can  be induced.  Natural spawning  is  preferred becaise repeated

  Kl"*S C,ah  bt °b,taired fr°m the Same brood  stock»  whereas wtth hormone
  injection,  the brood stock is sacrificed in obtaining gametes.

  7.15.2.2  It  should be  emphasized that  the  injection  and hatching schedules
  given below are to be used only as guidelines.  Response to the  hormone
  varies from stock  to stock and with temperature.   Time  to hatch  and percent
  viable hatch also  vary among stocks and  among batches of embryos  obtained
 from  the  same  stock, and  are dependent on temperature,  DO, and salinity
 The coordination of spawning and  hatching is further comp icated  by the'fact
 that,  even under the most ideal conditions,  embryos spawned over  a 24-h
 period may hatch over a  72-h  period.  Therefore,  it is  advisable  ^specially
 if  natural spawning is used)  to obtain fertilized  eggs over several dm to

                                       hatched
 7.15.2.3  Forced Spawning


 7.15.2.3.1   HCG is reconstituted with sterile saline or Ringer's solution
 inmediately before use.  The standard HCG  vial contains 1,000 IU to be
 reconstituted  in 10 ml of saline.  Freeze-dried HCG which comes wi?h
.premeasured and sterilized saline is the easiest to use.  Use of a 50 IU
 dose requires  injection of 0.05 ml of reconstituted hormone solution.
 Reconstituted HCG may be used for several weeks if kept in the refrigerator
Adapted from Hansen, et. a!., 1978.

-------
  7.15.2.3.2  Each female is injected intraperitoneally with 50 IU HCG on  two
  consecutive days, starting at least 10 days prior to the beginning of a
  test.  Two days following the second injection,  eggs are stripped from the
  females and mixed with sperm derived from excised macerated testes.   At
  least ten females and five males are used per test to ensure that there  is a
  sufficient number (400)  of viable embryos.

  7.15.2.3.3  HCG is injected into the peritoneal  cavity,  just below the skin
  using as small  a needle  as possible.   A  50  IU dose is recommended  for
  females approximately 27 mm in  standard  length.   A larger  or Siller dose
  may  be used for fish  which are  significantly  larger  or smaller than  ?7 ™
  s ouldtC±v fde  rd^S  °ne and  tw°' f-a?ersgewh?chS:?e1heridth   II "'
  should be  ready for stripping on days  4,  5, and  6.   Ripe females should show
  pronounced  abdominal .swelling,  and release at  least  a few eggs in response
  to a  gent e  squeeze.   Injected  females should  be  isolated from males   It
  S± •« Pf)l   "  f\Sh that are  to be  ™Jected are Maintained a^looc
  before  injection,  and  the  temperature raised to 25<>C on the day of the
  T irst  injsction*

  7.15.2.3.4  Prepare the testes  immediately before stripping the eqqs  from
  the females.  Remove the testes from three-to-five males.  The testes a?e

               r^ or9anS-al°n? ^ d°rsa1 midl1ne of ™e  a"dSmina  cav ty.
 fi,   mn<  „   'he.male " cut off and Pu"ed away from the rest  of the
 fish, most of the internal organs can be  pulled out of the body cavity
           ae^
 remove  the ovaries  if  all  the  ripe  eggs  do  not flow out freely?  Break up
 any clumps of  ripe  eggs  and  remove  clumps of ovarian tissue and underripe
 clear.   "* *"*  *™ Spherica1' «PProx1mately 1 mm in diameter, and lUost
 7.15.2.3.6  While being held over the dish containing the eggs, the testes
 are macerated  in a fold of NITEX* screen (250-500 urn mesh) dampened with
 seawater.  T e testes are then rinsed with seawater to remove the sperm from
 tissue, and the remaining sperm and testes are washed into the dish   Let
 the eggs and milt stand together for 10-15 min, swirling occasionally.

 7.15.2.3.7  Pour the contents of the dish into a beaker, and insert an
 a!rstone   Aerate gently,  such that the water moves slowly over?he eL
 and incubate at 25°C for 60-90 min.  After incubation,  wash the eggs on a
 Nitex screen and resuspend them in clean seawater.   Examine the eggs
 periodically under a dissecting microscope until  they are in the 2-8 cell
Irnf ;J  K6 StT 9J Wh1'Ch  H 1s eas1est to tel1  the developing embryos
from the abnormal  embryos  and unfertilized  eggs;  see Figure 1)    The eaas
can then be gently rolled  on  a Nitex screen and culled  (Paragraph  7 15 2 5  )
                                      42

-------
 7.15.2.4  Natural  Spawning

 7.15.2.4.1  Cultures of adult fish to be used for spawning are maintained at
 18-20°C until  embryos are required.   When embryos are required, raise the
 temperature to 25°C in the morning,  seven or eight days before the
 beginning  of a test.  That afternoon, transfer the adult fish (generally, at
 least  five females and three, males)  to a spawning chamber (approximately,
 20X35X22 cm high;  Hansen, et al.,  1978), which is a basket constructed of
 3-5  mm nylon mesh, made to. fit a 57-L (15 gal) aquarium.  Eggs will  fall
 through the bottom of the basket and onto a collecting screen (250-500 urn
 mesh)  below the basket.  Allow the embryos to collect for 24 h.  Embryos  are
 washed from the screen, checked for  viability, and placed in incubation
 dishes.  Replace the screens until a sufficient number of embryos have been
 collected.   One-to-three spawning  aquaria can be used to collect  the
 required number of embryos to run  a  toxicity test.   To help  keep  the embryos
 clean,  the adults  are fed while the  screens are removed.

 7.15.2.5  Incubation

 7.15.2.5.1   Four hours post-fertilization,  the embryos obtained by natural
 or forced  spawning are rolled gently with a finger  on a 250-500 urn nylon
 screen  to  remove excess fibers and tissue.   The embryos have adhesive
 threads  and  tend to  adhere to each other*   Gentle rolling  on the  screen
 facilitates  the culling process described below.

 7.15.2.5.2   Under  a  dissecting microscope,  separate and discard abnormal
 embryos  and  unfertilized  eggs.   While they  are checked,  the  embryos  are
 maintained  in  sea  water at 25°C.   The embryos  should  be in Stages  C-G,
 Figure  1.

 7.15.2.5.3   If  the test is prepared  with  four  replicates of  15  larvae at
 each of  six  treatments  {five  effluent concentrations  and a control),  and  the
 combined mortality of  eggs and  larvae prior to the  start of  the test  is less
 that 20%, approximately 400  viable embryos  are required  at this stage.

 7.15.2.5.4   Embryos  are demersal.  They  should  be aerated and  incubated at
 25°C, at a salinity  of  20-30  °/oo  and  a  14-h photoperiod.  The  embryos
 can be cultured  in either  a flow-through  or static  system, using aquaria  or
 crystallization  dishes.   However,  if  the  embryos  are  cultured  in dishes,  it
 is essential that  aeration and  daily  water  changes  be  provided, and the
 dishes be covered  to reduce evaporation that may cause  increased salinity.
 One-half to  three-quarters of  the  sea  water from the  culture vessels  can  be
 poured off and  the incubating  embryos  retained.   Embrybs cultured  in  this
manner should hatch  in  six or  seven days.

 7.15.2.5.5   At  48  h  post-fertilization, embryos are examined under a
microscope to determine development and survival.   Embryos should be  in
Stages I and J, Figure  1.  Discard dead embryos.  Approximately 360 viable
embryos are  required at this  stage.

-------
Figure -1.  Embryonic-development of sheepshead minnow (Cyprinodon
varieqatus): A. Mature unfertilized egg, showing attachment filaments and
nricropyTeT X33; B. Blastodisc fully developed; . C,D.  Blastodisc,  8 cells;
E. Blastoderm, 16 cells; F. Blastoderm, late cleavage stage; 0.  Blastoderm
with germ ring formed, embryonic shield developing; H. Blastoderm covers
over 3/4 of yolk, yolk noticeably constricted; I. Early embryo.   (From
Kuntz, 1916.)
                                   44

-------
         RM/-&S  Kw   Aax'f *fc-^ -V
Figure l   (Continued).   Embryonic
.      ,
                         45
                                    of

-------
 8.  EFFLUENT AND RECEIVING WATER COLLECTION, PRESERVATION, AND STORAGE

 8.1  See Section 8, Effluent and Receiving Water Sampling and Sample
 Handling.

 9.  CALIBRATION AND STANDARDIZATION

 9.1  See Section 4, Quality Assurance.

 10.  QUALITY CONTROL

 10.1  See Section 4, Quality Assurance.

 11.  TEST PROCEDURES

 11.1  TEST SOLUTIONS

 11.1.1   Surface Waters

 11.1.1.1   Surface water toxicity is determined  with  samples  used  directly  as
 collected.                                                              J

 11.1.2   Effluents

 11.1.2.1   The  selection of  the  effluent test concentrations  should be based
 on  the  objectives of the study.   One of two dilution factors, approximately
 0.3 or  0.5,  is  commonly used.   A dilution factor of approximately 0.3 allows
 Mn^"9 b*twee£ 100% and ]% eff1uent us™9 only five effluent concentrations
 (100%,  30%,  10%,  3%, and 1%).   This series of dilutions minimizes the level
 of  effort, but  because  of the wide  interval between test concentrations
 provides  poor test  precision  (see Section 4S Quality Assurance).  A dilution
 factor  of 0.5 provides  greater  precision, but requires several additional
 dilutions to span the  same range of effluent concentrations.  Improvements
 in  precision decline rapidly as  the dilution factor is increased
 beyond  0.5.  If 100  o/oo salinity MSB  is as a diluent, the maximum
 concentration of  effluent that  can be  tested will be 80% at 20 o/oo
 salinity and 70%  at  30  °/°o salinity.

 11.1.2.2  If the effluent is known or  suspected to be highly toxic, a lower
 range of effluent concentrations should be used, with a maximum
 concentration of  10%.   If a high rate of mortality is observed during the
 first l-to-2 h  of the test, additional  dilutions should be added at the
 lower range of  effluent concentrations.

 11.1.2.3  The volume of effluent required for daily renewal of three
replicates per concentration, each containing 750 mL of test solution,  is
approximately 5 L.  Prepare enough test solution (approximately 3000 ml)  at
each effluent concentration to provide  400 mL additional  volume for chemical
analyses (Table 1).

-------
   11.1.2.4   The  salinity of  effluent  and  receiving water  tests for  sheepshead
   minnows should be  between  20  °/oo and 30  o/oo.  If concurrent  effluent
   and receiving  water testing occurs, the effluent test salinity should
   closely approximate that of the receiving water test.   If an effluent is
   tested alone,  select a  salinity between 20 o/oo and 30  o/oo, whichever
   comes closest  to the salinity of the receiving waters.  Table  1 illustrates
   the quantities  of  effluent, artificial sea salts, hypersaline  brine, or
   seawater needed to prepare 3 L of test solution at each .treatment level for
   tests performed at 20 o/00 salinity.

   11.1.2.5  Approximately one hour before test initiation, warm a sufficient
  quantity of sample to 25 + 2°C to make the test solutions.

  11.1.2.6  Higher effluent concentrations (i.e., 10,  32,  and  100%)  may
  require aeration to maintain  adequate  dissolved oxygen  concentrations
  However,  if one solution is aerated, all  concentrations  must  be aerated
  Aerate effluent as  it  warms and  continue to  gently aerate test  solutions  in
  the  test chambers for  the duration of  the  test.

  11.1.2.7   Tests should  begin as  soon as  possible, preferably within  24 h
  after  sample collection.  If the persistence of the sample toxicity  is not
  known, the  maximum  holding  time following  retrieval of the sample  from the
  sampling device should  not  exceed 36 h for off-site toxicity studies.  In no
  case should the sample  be used in a  test more than 72 h  after sample
  collection.  Just prior  to  testing,  the temperature of the sample  should be
  adjusted to (25  + 2°C) and  maintained at that temperature until portions
  are added to the dilution water.

  11.1.3  Dilution Water

  11.1.3.1  Dilution water may be natural seawater (receiving  water)
 hypersaline brine prepared from natural seawater,  or  artificial  seawater
 prepared from FORTY  FATHOMS^ sea  salts  (see Section  7).   Other artificial
 sea salts may be used for culturing  sheepshead  minnows  and for the  larval
 survival  and growth  test if  the control  criteria for  acceptability  of test
 data  are  satisfied.

 11.2  START OF THE TEST

 11.2.1   If  the embryos have  been  incubating at 25^0, 30 o/oo salinity
 and a 14-h  photoperiod,  for  five-to-six days with aeration and daily  water
 renewals, approximately  24 h prior to hatching,"  the salinity of  the sea
 water in the  incubation  chamber may be reduced from 30 o/00 to the  test
 salinity, if  lower than  30 o/oo.  In  addition to maintaining good water
 quality, reducing the salinity and/or changing the water  may also help to
 initiate hatching over the next 24 h.  A few larvae may hatch 24 h  ahead of
 the majority.  Remove these  larvae and reserve them in a  separate dish,
 maintaining the  same culture conditions.  It is preferable to use only the
 larvae that hatch in the 24  h prior to starting the test.  However,  if
 sufficient numbers of larvae do not hatch within the  24-h period, the larvae
that hatch prior  to 24 h are added to the test organisms.   The test
organisms are then randomly selected for the test.

                                      47

-------
 11.2.2   Label  the  test  chambers  with  a  marking  pen and  use color  coded  tape  to
 identify each  treatment  and replicate.  A minimum of five effluent
 concentrations  and  a  control should be  used for each study.  Each treatment
 {including controls)  should have four  (minimum of three) replicates   For
 exposure chambers,  use  1000 ml beakers, non-toxic disposable plasticware  or
 method           th " SUmP area dS lllust"*d  ™ the inland silver side test

 11.2.3  Distribute the test solutions to the test chambers.

 11.2.4  The test is started by placing  larvae from the common pool,  one or two
 at a time  into each test chamber in sequential order,  until  each chambe?
 contains 15 {minimum of 10) larvae, for a total of 60 larvae  for each
^Kment lmnlmm f three replicates).  The amount  of water added  to the
chambers when transferring the larvae should be kept  to a minimum to avoid
unnecessary dilution of the test  concentrations.


larvae
             Chafflbers fflay be placed on a ]i9ht table <* facilitate counting the
 11.2.6  Randomize the position of the test chambers  at  the  beginning of the
 thf tlcf APpendlx>t   Ma|ntain the chambers in this  configuration  throughout
 the test.  Preparation of a position  chart may be  helpful!            ^uynuuc

 11.3   LIGHT,  PHOTOPERIOD,  SALINITY,  AND  TEMPERATURE

 11.3.1   The light quality and  intensity should be  at ambient  laboratory
  evels,  which  is  approximately 10-20  u£/m2/s,  or 50  to  100  foot  candles
 !h   ?KW1* ?  Phot°Peri?d  of  14  h  light  and  10 h  darkness.   The test  salinity
 should  be in the  range of 20 to 30  o/00 to accommodate  receiving waters that
 may fall  within this  range.  The  salinity should vary by no more than
 + 2 o/oo- among  the chambers on a  given day.   If effluent and  receiving  water
 tests are conducted concurrently,  the salinities of  these tests  should  be
 25  + 2°C      ^^ temperature 1n  the test chambers should be maintained at


 11.4  DISSOLVED OXYGEN CONCENTRATION  (DO)

 11.4.1  Aeration may affect the toxicity  of effluents and should be used only
as  a last resort to maintain a satisfactory DO.  The DO concentrations  should
be measured on new solutions at the start  of the test (Day 0) and before daily
renewal of  test solutions on subsequent days.   The DO should not fall below
40% saturation (see Section 8).  If it is  necessary to aerate, all  treatments
?nn u"^control should be aerated.  The aeration rate should not exceed
100 bubbles/mm, using a pipet with an orifice of approximately 1.5 mm, such
as a 1-mL Kimax serological pipet, No. 37033,  or equivalent.  Care  should be
taken to ensure that turbulence resulting  from aeration  does not cause undue
stress on the fish.
                                       48

-------
 11.5  FEEDING
  11.5.1  Sheepshead minnow  larvae are fed newly-hatched
    ,             ..      s a







                                                   •      -

11-6  DAILY CLEANING OF TEST CHAMBERS

                                             r ,   r «
                                    '
                                           «
      „ a, >,,»„„ „„ „, r,tr,e,ea,a ™,    j th
,nc,«nc, of r-»,.l of 11,, !.„,. fro, tl» uit cha.Mr,
                                                 t,

                                                   «    «

-------
H.7  TEST SOLUTION
                         RENEWAL
                                    nnh
     tox-rcity  studies, fresh effl.»n?   2  "* test  chambers.  For

     toxicity  tests should be coneectedndanCveiV1'n9 Water ^


     '^            «  £ ^l y-n-'«.n°CSI.tS2
     .  P
                                         th. test, warra the
                                                                     (jf
  11.8  OBSERVATIONS DURING THE TEST



  "•8-1    Routine Chenica! and Physica,  Observations




                                      .easure.ents are made and  recorded
 control.  The PH is wasureV"^^



 11-8.2  Routine Biological Observations


IK8.2.1 The number  of

        7),  and  the
                                                          recorded dai,y


i ("OBC.2  Ppotepf" f*h^ i
                                    50

-------
TABLE 1.

                                       Solutions  To  Be  Combined
                                       .                  -  •  	
                                                    Volume  of  Diluent
                                                   Seawater  (20 o/00)
 Effluent
 Solution


   T


   2


   3


   4


   5


Control
             1001.2


              32


              10


               3.2


               1.0


             0.0
 5100 ml


 1700 ml Solution  1


 1700 mL Solution 2


1700 ml Solution 3


1700 ml Solution 4
 3400 nil


 3400 ml


 3400 ml


3400 ml


3400 ml
                                         °hemiCa1  anal*sis  (*°tal  of
 3400 ml) for the control and
 effluent dilution facto  of 03  and
 sa inity.  A sufficient initi 1 volume  5 00 mi
 adjustmg the salinity to the desired level   Tn
 TS adjusted by adding artificial  ea salts
 preparing a serial  dilution i,<:inn ?n o/l,
 hypersaline brine  or ar ?fic  1

              s
                                                         °f
                                                             1S PrePared by

                                             the 10M        >*the 5al1nity
                                                  ,0/i effluent, and

                                                   "atUral seawater'

                                  51

-------
    H.10  TERMINATION OF THE TEST
              ss
charter are  counted an
                    '"
                             ed ate

                             «>""»"
                                              days °
                                             "
                                                             " 6ach test
                                                                          .7
   a f° «f°^^                             "  •"•
   ArteBla and debris to be rinsed awav   »?„« f^ ?he 'arvae and a1'ow
   to wash away salts that »ight1oRut2          """?  "h de1o"<«d water
              by pacing th.'fn an
                                                    to
 weighing boats (one  per rep   ate)
 record the weights (Figure 8)
                                                       °f Sma11
                                                   nearest °-01 "B.  and
  .   ng all weighing boats containina dH»H
"eight to determine the dry weight $ lar«e
weights (Figure 8).   For each t«t cLmh2
the number of larvae surviving In the test
individual dry weight, and record (Fiaurf «?
sheet (Figure 9)                           K
statistically
                                                              "eare«
                                                   ha"d  S^tract the
                                                "^/epTlcate.  Record  the
                                                   *he/inal ^ry weight  by
                                              r    ^determine the average
                                              ComPlete the summary data
 12.  SUMMARY OF TEST CONDITIONS


 12.1  A summary of test conditions  is listed in Table  2.

 13-  ACCEPTABILITY 'OF TEST RESULTS
                                            a                of control
unpreserved control  larvae is eouai %«       average dry weight of
average dry weight of pre'se ved'co t ol  Lr'va'e ?^ .^V^0 ^ °r (3) the
0.50 «g.   The  above minimum weight I prism ^ that tM0t°*ori.9r?ater than
the start  of the test is less than o? equal to 24 h   9     the 1arvae at

-------
TABLE 2.  SUMMARY OF RECOMMENDED TEST CONDITIONS FOR SHEEPSHEAD MINNOW
(CYPRINODON VARIEGATUS) LARVAL SURVIVAL AND GROWTH TEST
  1.  Test type:
  2.  Salinity:
  3.  Temperature:
  4.  Light quality:
  5.  Light intensity:   ;.J

  6.  Photoperiod:
  7.  Test chamber size:
  8.  Test solution volume
  9.   Renewal  of  test
       concentrations:
 10.   Age  of test organisms
 11.  Larvae/test chamber
 12.  Replicate
      chambers/concentration
 13.  Source of food:
14.  Feeding regime:
15.  Cleaning:

16.  Aeration:
 Static renewal
 20 o/oo to 32 o/oo + 2 o/00
 25 + 20C
 Ambient laboratory illumination
 10-20 uE/m2/s (50-100 ft-c) (ambient
 lab levels)
 14 h light, 10 h darkness
 300 mL - 1  L beakers or equivalent
 250 - 750 mL/replicate (loading and
 DO restrictions  must be met)
 Daily
 Newly hatched  larvae (less  than
 24 h  old)
 15 larvae/chamber   (minimum of  10)

 4  (minimum  of  3)
 Newly  hatched  Artemia)  nauplii
 (less  than  24  h old)
 Feed once a day 0.10 g  wet weight
 Artemia nauplii per replicate on
 Days 0-2; feed 0.15 g wet weight
 Artemia nauplii per replicate on
 Days 3-6
 Siphon daily,  immediately before test
 solution renewal
 None, unless DO falls below
60% of saturation, then aerate all
chambers.  Rate should be
 less than 100 bubbles/min.

-------
 17.  Dilution water:
 18.

 19.

20.

21.
 Effluent  concentrations

 Dilution  factor:

Test duration:

Effects measured:
TABLE 2. CONTINUED
      "" '       --   	

      Uncontaminated source of natural
      seawater, or hypersaline brine or
      artificial seawater mixed with
      deionized water

      5  and a control

      Approximately 0.3  or  0.5

      7 days

     Survival and growth (weight)
                                 54

-------
    H.  DATA ANALYSIS


    14.1   GENERAL
   CM ne,

   using a'hypothesis  tes  approach  usn'      rwt?>  •"  obtained

   or Steel's Many-one Rank TeTfSteeU ?959  miL PCSS?fUrec(Dunnrtt.  I9«)

   for examples of the manual computat ons  DroarJm i^t? 'K   S!e  the  APPend«
   data  input and program output.           program Imings,  and examples of




                                tte Tests  a^"1'^  be US6d Wlth a k"°»^d9e
  and homogeneity of variance  are  included  ^n ^l"9*^-  Tests for normality
  a statistician  is  recommended  lor analyst J^e ApPendlx-  T"e assistance  of
  statistics.                       analysts who are not proficient in



  14.2 EXAMPLE OF ANALYSIS OF SHEEPHEAD MINNOW SURVIVAL  DATA
 surviving  in each test or control
 for  the  estimation of the ml ™ nd
 the  LCI, LC5, LC10 and LC50 end points

            =
                                                            .
                                                    Proportion of animals


                                                              8re performed
                                                              est1>nation of
nonparametric  test,  Steel's
                                                           21
                                                           or a
                                     r
Many-one Rank Test,  is  used  to  determine thl JS?r °n5a,r^tr1c test> Steel's
the assumptions of Dunnett's Procedure a~ t?  ti."1" ^°EC 6nd points-   If
by the parametric  procedure  Hr°Cedure are met> the end points are estimated
                                    55

-------
      TABLE 3. SUMMARY OF SURVIVAL AND GROWTH DATA FOR  SHEEPSHEAD  MINNOW
               LARVAE EXPOSED TO AN EFFLUENT FOR SEVEN
Effl.
Cone.
(%)
: 0
0.32
1.0
3.2
10.0
32.0
Proportion of
Survival in Replicate
Chambers

1
1
1
1
0
0
A
.0
.0
.0
.0
.8
.0
B
1.0
1.0
1.0
1.0
0.8
0.0
c
1.0
0.9
1.0
1.0
0.7
0.0

1
1
1
0
0
0
D
.0
.0
.0
.8
.6
.0
Mean
Prop.
Surv
1
0
1
0
0
0,
.00
.98
.00
.95
.73
.00
Ave Dry Wgt (mg) In
Replicate Chambers
ABC
1.29 1.32 1.59
1.27 1.00 1.08
1.32 1.37 1.35
1.29 1.33 1.20
V 0.78 0.70 0.66
—
D
1.27
0.97
1.34
1.17
0.77
—
Mean
Dry Wgt
(mg)
1.368
K080
1.345
1.248
0.728
—
Vour replicates of 10 larvae each.
                                     56

-------
       STATISTICAL ANALYSIS OF  SHEEPSHEAD MINNOW LARVAL
                     SURVIVAL AND GROWTH TEST

                            SURVIVAL
        f.
      PROBIT
      ANALYSIS

                             SURVIVAL  DATA
                         PROPORTION SURVIVING
               ARCSIN
           TRANSFORMATION
                         HAPIRO-HILKS
             NORMAL DISTRIBUTION


                        ^BARTLETT'S TEST

HOMOGENEOUS VARIANCE
       NO
                                            NON-NORMAL DISTRIBUTION
                                  HETEROGENEOUS
                                     VARIANCE
              EQUAL NUMBER OF
                REPLICATES?
                YES
   T-TEST WITH
   BONFERRONI
  [_ ADJUSTMENT
       T ""
   T
                     EQUAL NUMBER OF
                      REPLICATES?
                                       NO
                                      YES
OUNNETT'S
  TEST
STEEL'S MANY-ONE
   HANK TEST
  WILCOXON RANK SUM
      TEST WITH
BONFERRONI ADJUSTMENT
                         ENDPOINT  ESTIMATES
                              NOEC. LOEC
        Figure 2. Flow chart for statistical  analysis  of  sheepshead
                  minnow larval  survival  data.
                                57

-------
14.3.3  If unequal numbers of replicates occur among the concentration
levels tested, there are parametric and nonparametric alternative analyses.
The parametric analysis is the Bonferroni t-test.  The Wilcoxon Rank Sum
Test with the Bonferroni adjustment is the nonparametric alternative.  For
detailed information on the Bonferroni adjustment see the Appendix.

14.3.4  Probit Analysis (Finney, 1971) is used to estimate the concentration
that causes a specified percent decrease in survival from the control.  In
this analysis, the total mortality data from all test replicates at a given
concentration are combined.

14.3.5  Example of Analysis of Survival Data

14.3.5.1  This example uses the survival  data from the Sheepshead Minnow
Larval Survival and Growth Test.  The proportion surviving in each replicate
must first be transformed by the arc sine square root transformation
procedure described in Appendix B.   The raw and transformed data, means and
standard deviations of the transformed observations at each effluent
concentration and control  are listed in Table 4.  A plot of the survival
proportions is provided in Figure 3.  Since there was 100% mortality in all
four replicates for the 32% concentration,  it was not included in the
statistical analysis and was considered a qualitative mortality effect.
                  TABLE 4.  SHEEPSHEAD MINNOW SURVIVAL DATA
             Replicate     Control
0.32
Effluent Concentration!^

                       10.0
1.0
3.2

RAW


ARC SINE
TRANSFORMED


ME AN (Til
Si2
i
A
B
C
D
A
B
C
D



1.0
1.0
1.0
1.0
1.412
1.412
1.412
1.412
1.412
0.0
1
1.0
1.0
0.9
1.0
1.412
K412
1.249
1.412
1.371
0.007
2
l.Q
1.0
1.0
1.0
1.412
1.412
1.412
1.412
1.412
0.0
-j .'
1.0
1.0
1.0
0.8
1.412
U412
1.412
- 1.107
1.336
0.023
J 4
0.8
0.8
0.7
0.6
1.107
1.107
0.991
0.886 .
1.023
0.011
5
                                      58

-------
UJ
o

§
O
 (X
 o
 uJ
  O
  UJ
  o
  o
S-
o
Q-
O
S~
Q-
                                                                                               ro
                                                                                               3
                                                                                               VJ
                                                                                               fO
                                                                                               a>
                                                                                                o


                                                                                                o

                                                                                                Q.

                                                                                                 •
                                                                                                ro

                                                                                                
                                                                                                -r—
                                                                                                U-
                                   NOlidOdOdd
                                                   59

-------
   14.2.6  Test for Normality                               ~   ""

   14.2.6.1  The first step of the test for normality is to center the
   observations by subtracting the mean of all observations within a
   concentration from each observation in that concentration  The centered
   observations are summarized in Table 5.            '<*<•'"".  me centered

             TABLE 5.  CENTERED OBSERVATIONS FOR SHAPIRO-WILKS EXAMPLE
   Replicate
       A
       B
       C
       D
                                                  Concentration  (%)
Control
0.32
                                                 1.0
                                   3.2
                                                                    10.0
0.0
0.0
0.0
0.0
0.041
0.041
-0.122
0.041
0.0
0.0
0.0
0.0
0.076
0.076
0.076
-0.229
0.084
0.084
-0.032
-0.137
  14.2.6.2   Calculate  the  denominator,  0,  of  the  statistic:

                           n
                       D  = £  (Xt - x)2
                          i = l

     Where   X-j = the ith centered observation         .-;&"
             X  = the overall  mean of the centered observations
             n  - the total  number of centered observations

 14.2.6.3  For this set of data:     n = 20     ^-v


                                    * s J_ (-0-001)  =  0.000
                                         20

                                   D -  0.1236

 14.2.6.4  Order the centered  observations  from smallest to largest

               XH) -X<2) -  ... - X(n)

where x(|) denotes the ith ordered observation.  The ordered
observations for this example are listed in Table 6.        °
                                    60

-------
    TABLE 6.
             ORDERED CENTERED OBSERVATIONS FOR THE SHAPIRO-WILKS EXAMPLE
1
2
3
4
5
6
7
8
9
10
-0.229
-0.137
-0.122
-0.032
0.0
0.0
0.0
0.0
0.0
0.0
11
12
13
14
15
16
17
18
19
20
0 0
0.0
0.041
0.041
0.041
0.076
0.076
0.076
0.084
0.084
14.2.6.6  Compute  the  test  statistic, W,  as follows:

                   1    k
               W = ~ l.Zfi
                                              ]2
                          (0-3178)2 , 0.8171
TABLE 7.  COEFFICIENTS AND DIFFERENCES  FOR  SHAPIRO-WILKS  EXAMPLE
         1
         2
         3
         4
         5
         6
         7
        8
        9
       10
               0.4734
               0.3211
               0.2565
               0.2085
               0.1686
               0.1334
               0.1013
               0.0711
               0.0422
              0.0140
 0.313
 0.221
 0.198
 0.108
 0.076
 0.041
 0.041
 0.041
0.0
0.0
      - xO)
      - X<2)
      - X<3)
XH6) . X(5)
X03) . X(8)
                                   61

-------



      «
  14.2.7  Steel's Many-One Rank Test
s£s
            r,n  to

      10.

TABLE  8.
                                 -

                                        .
                                                         CONCENTRATION
               Rank
               1
               5
               5
               5
               5
              5
              5
              5
                                Transformed
                                Proportion
                                Surviving
                                 1.249
                                 1.412
                                 1.412
                                 1.412
                                 1.412
                                 1.412
                                 1.412
                                 1.412
                                                      Effluent
                                                      Concentration
 0.32
 0.32
 0.32
 0.32
 Control
 Control
Control
Control
                                62

-------
Repli-
cate
A
B
C
D
Control
1.412(5,4,5,3.5)
1.412(5,4,5,3.5}
1.412(5,4,5,3.5}
1.412(5,4,5,3.5)
                             TABLE 9. TABLE OF RANKS
                                     IffjuentConcentration  (%)
                                  0.32          i.o         3.2
1.412(5)
1.412(5)
1.249(1)
1.412(5)
1.412(4}
1.412(4)
1.412(4)
1.412(4)
            10.0
                                                       1.412(5)  1.107(3.5}
                                                       1.412(5)  1.107(3.5)
                                                       1.412(5)  0.991(2}
                                                       1.107(1)  0.886(1)
  ?•
                             TABLE  10.  RANK SUMS
                     Effluent  Concentration  (%)
Rank Sum
                                0.32
                                1.0
                                3.2
                               10.0
   16
   16
   16
   10
 i™%    °r     °W  WhlCl;  the  Surv1val  would  be  "leered  significantly
 lower  than  the control.  At  a  significance  level of 0.05, the minimum
 rank SUB. in a  test with four concentrations (excluding the control and
 four replicates is 10 (See Table 5,  Appendix  E).           control) and

 tn'th!'!  -t-nC? th? rankLsum for the 10» effluent concentration is equal
 to the critical value, the proportion surviving in the 10% concentration
 is considered signlf cantly  less than that in the control.  S nee no
other rank sums are less than or equal  to the critical value  no other
concentrations have significantly lower proportion surviv ng than thT

                                           "Sumed t0  **&«£ lOJ,
14.2.8  Probit Analysis

                   useMor  ^e  Probit  analysis  is  summarized  in
                             "n
                                    63

-------
14.2.8.2  For this example, since there is 100% survival in the controls,
there is no need to adjust for control mortality.  The test for heterogeneity
was significant, thus confidence limits for the LC values could not be
calculated.  Probit analysis does not appear appropriate in this case.
                       TABLE  11.   DATA  FOR  PROBIT ANALYSIS
                                       Effluent Concentration (%)
                       Control
           0.32
       1.0
       3.2
       10,0
       32.0
Number Dead
Number Exposed
 0
40
 1
40
 0
40
 2
40
11
                                       64

-------
TABLE  12.  OUTPUT  FROM EPA PROBIT ANALYSIS  PROGRAM, VERSION  1.3,
                  USED FOR CALCULATING EC VALUES
Probit Analysis of Sheepshead Minnow Larval Survival Data
    Cone.

    0.3200
    1.0000
    3.2000
   10.0000
   32.0000
 Number
Exposed

    40
    40
    40
    40
    40
Number
Resp,

    1
    0
    2
   11
   40
 Observed
Proportion
Responding

  0.0250
  0.0000
  0.0500
  0.2750
  1.0000
 Adjusted
Proportion
Responding

  0.0250
  0.0000
  0.0500
  0.2750
  1.0000
Predicted
Proportion
Responding

  0.0000
  0.0034
  0.0789
  0.4437
  0.8761
Chi - Square Heterogeneity =  737.067
*                        WARNING                              *
*                                                             *
*   Significant heterogeneity exists.   The results reported   *
*   for this data set may not be valid.  The results should   *
*   be interpreted with appropriate caution.                  *
***************************************************************
**************!

*
                                     :**************************
                           NOTE
Mu
Sigma
      Slope not significantly different from zero.
      EC fiducial limits cannot be computed.
    1,055150
    0.389343
Parameter
    Estimate
    Std. Err.
             95% Confidence Limits
Intercept
Slope
    2.289923
    2.568429
    5.958668
    5.498257
       (   -16.670557,
       (   ~14.927025,
           21.250403)
           20.063881)
Theoretical Spontaneous Response Rate - 0.0000
      Estimated EC Values and Confidence Limits
Point

EC 1.00
EC 5.00
EC10.00
EC15.00
EC50.00
EC85.00
EC90.00
EC95.00
EC99.00
       Cone.

        1.4107
        2.5985
        3.5989
        4.4837
       11.3540
       28.7518
       35.8205
       49.6113
       91,3858
              Lower "      Upper
            95% Confidence Limits

-------
probit Analysis of Sheepshead Minnow Larval Survival Data

        PLOT  OF ADJUSTED PROBITS AND PREDICTED REGRESSION LINE
Probit
   10 +
     5 +
     4 +
     3+0
     2 +
     0-K)
       EC01
                      EC10     EC25
                                          4	

                                         EC50
EC75     EC90
                       „ H	I M

                       EC99
     Figure 4. Plot of adjusted probits and  predicted regression  line
                            from EPA Probit Program

-------
14.3  EXAMPLE OF ANALYSIS OF SHEEPSHEAD MINNOW GROWTH DATA

14.3.1  Formal statistical analysis of the growth data is outlined in
Figure 5.  The response used in the statistical analysis is mean weight
per replicate.  Concentrations above the NOEC for survival are excluded
from the growth analysis.

14.3.2  The statistical analysis consists of a parametric test, Dunnett's
Procedure, and a non-parametric test, Steel's Many-one Rank Test. The
underlying assumptions of the Dunnett's Procedure, normality and
homogeneity of variance, are formally tested.  The test for normality is
the Shapiro-Wilks Test and Bartlett's Test is used to test for
homogeneity of variance.  If either of these tests fail, the
non-parametric test, Steels' Many-one Rank Test, is used to determine the
NOEC and LOEC end points.  If the assumptions of Dunnett's Procedure are
met, the end points are determined by the parametric test.

14.3.3  Additionally, if unequal numbers of replicates occur among the
concentration levels tested there are parametric and non-parametric
alternative analyses.  The parametric analysis is the Bonferroni t-test.
The Wilcoxon Rank Sum Test with the Bonferroni adjustment is the
non-parametric alternative.  For detailed information on the Bonferroni
adjustment, see the Appendix.

14.3.5  The data, mean and standard deviation of the observations at each
concentration including the control are listed in Table 13.  A plot of
the mean weights for each treatment is provided  in Fig. 6.  Since there
is no survival in the 32% concentration, it is not considered In the
growth analysis.  Additionally, since there is significant mortality in
the 10% effluent concentration, its effect on growth is not considered.
                  TABLE 13.  .SHEEPSHEAD MINNOW GROWTH DATA
                                   Effluent Concentration  (%)
Replicate    Control
0.32
1.0
3.2   10.0
32.0
A
B
C
D
Mean(Ti)
Si2
i
i
1.29
1.32
1.59
1.27
1.37
0.0224
1
1,27
0.998
1.08
0.97
1.08
0.0183
2
1.32
1.37
1.35
1.34
1.34
0.0004
3
1.29 -
1.33 -
1.20 -
1.17 -,
1.25 -
0.0056 -
4 5
.
-
. -

-
_
6
                                     67

-------
STATISTICAL ANALYSIS OF SHEEPSHEAD MINNOW LARVAL
SURVIVAL AND GROWTH TEST
GROWTH
(EXC
HOMOGENEOUS
NO
"
T-TEST WI
BONFERROh
ADJUSTMEf'


GROWTH DATA
MEAN HEIGHT
'LUDING CONCENTRATIONS ABOVE NOEC FOR SURV

'•-• •-•-- SH
I MOM- NDF1MJ

AHlHU^nlLlvb 1 La I
NORMAL DISTRIBUTION 1


VARIANCE
V
BARTLETT'S TEST 	 ** nc
V
_ EQUAL NUMBER OF EQUAL NUMBER OF
REPLICATES? REPLICATES?
YES 1
IH DUNNETT'S
U TEST
JT ltb'


1 YES
1 STEEL'S MANY-ONE WILC2
1 RANK TEST BONFERR

~T
ENDPOINT ESTIMATES
NOEC, LOEC
IVAL)
\L DISTRIBUTION
TEROGENEOUS
VARIANCE
NO
V
XON RANK SUM
EST WITH
ONI ADJUSTMENT




Figure 5. Flow chart for statistical  analysis of sheepshead minnow
                        larval growth data.
                                68

-------
                                                                       o
                                                                       c
                                                                       cr
                                                                      -,—
                                                                       E
                                                                      JT
                                                                       */»
                                                                       Q.
                                                                      (0 0>
                                                                     ID *J
                                                                     -»->
                                                                     -E -
                                                                      O)
                                                                      CU  S-
                                                                      S  en
                                                                      cr -o
                                                                      «3  E
                                                                      O>  rtJ
                                                                     o >

                                                                     -M >
                                                                     O i.
                                                                    -VO

                                                                     OJ
                                                                     S-
                                                                     ZI
                                                                     CO
                                                                    -r-
                                                                    ti.
     o»   oq
*•    o   o   d   6   d
(3H) 1H3I3/W NV3n
04   ^
6   6


-------
   14.3.6  Test for Normality

                                                     " 'center
          TABLE  14.  CENTERED OBSERVATIONS FOR SHAPRIO-WILKS
                                                             EXAMPLE
14.3.6.2  Calculate the denominator, D, of the
                                               test statistic:
    Where xt - the ith centered observation
          X  = the overall mean of the centered  observations
          n  = the total number of centered observations
For this set of data,
                               n -  16


                               X - JJ-0.006) = 0.000
                                    16
                               D - 0.1402
14.3.6.3  Order  the centered observations from smallest to  largest

                             - ... . X{n)
         In
                                           These  ordered observations
                                   70

-------
     TABLE 15.
               ORDERED CENTERED OBSERVATIONS FOR SHAPIRO-WILKS EXAMPLE
" '" ' ' • • .—
1
2
3
4
5
6
7
8
	 	 • —
-o.n
-0.10
-0.08
-0,08
-0.08
-0.05
-0.05
-0.02
9
10
11
12
13
14
75
16
-o.oo
0.00
0.01
0.03
0.04
0.08
0.19
0.99
n/Z;  For the data in
listed in Table 16.
                                                     * observations, n,
                                               - «     ™1«*™y
                                           ' k ~ 8-  The a1 values are
 14.3.6.5 Compute the  test  statistic,  w,  as follows

               W = D  Ml*1  (x(n"1+1)  -  x(f>)  ]2

the differences  x(n-1+l) . x(i) are ,-sted  fn ^

For this set of data:


                     W * -0^402 (0'3512)2 = °'880
 TABLE  16.   COEFFICIENTS AND DIFFERENCES FOR SHAPIRO
                                                       WILKS  EXAMPLE
1
2
3
4
5
6
7
8
0.5056
0.3290
0.2521
0.1939
0.1447
0.1005
0.0593
0.0196
                               0.33
                               0.29
                               0.16
                               0.12
                               0.11
                               0.06
                               0.05
                               0.02
                                                       - X(2)
                                                       - X<3)
                                                 X(12)
- X(5)
- X(6)
                                                      - X(8)
                                  71

-------
 14.3.6.6  The decision rule for this test is to compare W with the
 critical value found in Table 6, Appendix B.  If the computed W is less
 than the critical  value,  conclude that the data are not normally
 distributed.   For  this example, the critical value at a significance
 level  of 0.01 and  16 observations (n)  is  0.880.  Since W = 0.876 is
 greater  than  the critical  value, the conclusion of the test is that  the
 data are normally  distributed.

 14.3.7 Test for Homogeneity of  Variance

 14.3.7.1  The test used to examine whether the  variation  in mean dry
 weight is  the same across  all effluent concentrations  including the
 control, is Bartlett's Test (Snedecor  and Cochran,  1980).   The test
 statistic  is  as follows:
                                  p
                            -,
                [  ( Z Vj)  In $2 - I Vj  In S
           B =
Where Vj =
      p  =
                 degrees of freedom for each effluent concen
                 tration and control, Vj =  (n-j -  1)

                 number of levels of effluent concentration
                 including the control
          $2 =
                      P
                      t V
          C  = 1 + ( 3{p-l)H [ t 1/Vj - ( I
                                1*1      ,  i=l

         Where:
                 In =
                 i  - 1, 2, ..., p where p is the number of concentrations
                            including the control
               n-j = the number of replicates for concentration i.

14.3.7.2 For the data in this example, (See Table 13} all effluent
concentrations including the control have the same number of replicates
(nj = 4 for all i).  Thus, Vj => 3 for all i.
                                    n

-------
   '4-3.7.3  BarUetfs statijt,c ,
             9-049/1.139
                                3    ln(Si,2]/Km
                                1-1
                          - 3(-20.809)]/M39
          =  7.945
different.             ]1'3^'  concTude that the variances are noT tha"

'4-3.8  Dunnett's Procedure
                         TABLE 17.   ANOVA TABLE
   Source
                df
Sum of Squares
     (SSJ
                                                    Squ-are(MS)
                                                   (SS/dfJ
    SSB


    SSW
                                               SB = SSB/{p-7)
                                                z
                                                W = SSW/fN-pJ
      SST =  ,    /^
          =  SST  -  SSB
                                     Between Sum of Square
           Total Sum of Squares
                                     ithin Sum of Squares

-------
               G  - the grand total  of all  sample observations,  G .= I T-

               T7-  = the total of  the replicate  measurements  for
                    concentration  "i"
              YIJ  = the jth observation  for  concentration  "i"  (represents
                    the mean  dry weight  of the  fish for effluent
                    concentration  i  in test  chamber j)

  14.3.8.2  For the data in  this example:
 "1  =
 N   =

 T2  =

 T4 =

 P   _
           n2 = n3 = n4 = 4
           16
               + Y]2 + Y]3 + Y14 = 5.47
               + Y22 + Y23 + Y24 - 4.32
               + Y32 + Y33 + Y34 = 5.38
                 Y42 + Y43 + Y44 = 4.99

                T2 + T3 +  T4 = 20.16
     SSB =
             _(102.43) - (20.158)2  1 0
            4                16
           P    ni
     SST  =  I    £ Yij2  -  G2/N
         i-1  >1   J
        = 25.74   -  (20.16)2  = Oa34
                       16

    SSW = SST - SSB = 0.34 - 0.21 - 0.13
    Sfi2 = SSB/p-1 = 0.21/4-1 = 0.07
    Sw2 - SSW/N-p = 0.13/16-4 = 0.01

14.3.8.3 Summarize these calculations in the ANOVA table (Table 18)

             TABLE 18.   ANOVA TABLE  FOR  DUNNETT'S PROCEDURE  EXAMPLE
Source
df
                             Sum  of  Squares
                                  (SS)
                                              Mean Square(MS)
                                                  (SS/df)
   Total
                                      74

-------
   14..3.8.4  To perform the individual comparisons, calculate the t
   statistic for each concentration, and control combination as follows

                                      ( Yi - Yi )
  Where Yi  = mean dry weight for effluent concentration 1
        T]  = mean dry weight for the control
        SW  = square root of within mean  sqaure
        n]  = number of replicates  for control
        nj  - number of replicates  for concentration  i.

  14.3.8.5  Table  19 includes  the calculated t  values for  each
  concentration  and control  combination.   In this example, comparing the
  0.32%  concentration  with the control  the  calculation is  as follows-
                              (1.37  - 1.08 )
                                                  = 4.10
                        [ 0.10  V	(V4) + (1/4)  ]
                        TABLE  19.   CALCULATED T-VALUES
            Effluent Concentration^)
0.32
1.0
3.2
2
3
4
4.10
0.42
1.70
 14.3.8.6  Since the purpose of this test  is to detect a significant
 reduction  in mean weight, a  (one-sided)  test is appropriate.  The
 critical  value for this one-sided test is found in Table 5, Appendix C.
 For an overall alpha level of 0.05,  12 degrees of freedom for error and
 three concentrations (excluding the control) the critical value is 2 29
 The mean weight for concentration "i" is considered significantly less
 than the mean weight for the control if tj is greater than the critical
 value.  Since tg is greater than 2.29, the 0.32% concentration has
 significantly lower growth than the control.  However,  the 1.0% and 3.2?
concentrations do not exhibit this effect.  Hence the NOEC and the LOEC
for growth can not be calculated.
                                    75

-------

                           SW
                                            (17nT
Where  d

       ;>
       n
            = the critical value for the Dunnett's procedure

            = the square root of the within mean square

            = the common number of replicates at each  concentration

            - th*   3KSumel  equ?]  rePH«t1on at each  concentration
            - the number of  replicates  in  the control.
  14.3.8.8   In  this  example:
                    MSD = 2.29  (o.io)  v

                        = 2.29  (0.10)(0.707)
                        = 0.16
                                                                        can be



 14.3.8.10  This represents a 12% reduction in mean  weight  from the control.


 15.   PRECISION AND ACCURACY


 15.1   PRECISION



 15.1  1   Data  on the  single  laboratory precision of the sheepshead minnow

 larval  survival  and  growth  test  using FORTY FATHOMSR art liclafseawater and
 natural  seawater,  and rnnnor *t,i *»+**** MJ.«..« ^..__ ,    M:: .   ^aw^er and

 toxicants, are  listed

 tests was very  good.



 15.1.1  Data from a study of multilaboratory test precision, involvina a total

of seven tests by four participating laboratories,  are listed in Table ?4

InL^rat?neS reP?rted ver> similar results, indicating  good
interlaboratory precision.                                y  y


15.2  ACCURACY



15.2.1   The  accuracy of  toxicity tests cannot  be determined.
                                      76

-------
  TABLE 20. SINGLE LABORATORY PRECISION OF THE SHEEPSHEAD MINNOW
            (CYPRINIDON VARIEGATUS) LARVAL SURVIVAL AND GROWTH TEST
            PERFORMED IN FORTY FATHOMS* ARTIFICIAL SEAWATER, USING
            LARVAE FROM FISH MAINTAINED AND SPAWNED IN FORTY FATHOMS*
            ARTIFICIAL SEAWATER, AND COPPER AS A REFERENCE
            TOXICANT!,2,3,4,5
Survival
Test

1
2


5
6
7
8
i.
NOEC
(ug/L)

50
50
50
50
50
50
100
100
LOEC
(ug/L)

100
100
100
100
100
100
200
200
Growth
NOEC
(ug/L)

SE
• 50
«- 50
50
50
< 50
50
50
LOEC
(ug/L)

— .... . i ..I.,. ., tl „.
SE
50
50
SE
SE
50
100
100
Most
Sensitive
End Point
" 	 •" -ni in •, - |, ,
c
6
G


(2
G
G
 •Tests  performed by Donald J.  Klemm,  Aquatic Biology ^
  Newtown  Facility,  Environmental  Monitoring and Support
  Cincinnati.

 ZA11  tests were  performed  using Forty Fathoms*  synthetic  seawater.
  Three  replicate exposure  chambers, each with 15  larvae,  were  used  for
  the  control and each copper concentration.  Copper  concentrations
  used in  Tests 1-6 were: 50, 100, 200, 400,  and 800  ug/L.  Copper
  concentrations  in Tests 7-8 were: 25, 50,  100, 200  and 400 ug/L.
3Adults collected in the field.

4SE = Survival effects.  Growth data at these toxicant concentrations
 were^disregarded because there was a significant reduction in
 survival,

5For a discussion of the precision of data from chronic toxicitv
 tests see Section 4, Quality Assurance.
                                  77

-------
   TABLE 21
 SINGLE  LABORATORY PRECISION OF THE SHEEPSHEAD MINNOW

 (CYPRINIDON VARIEGATUS) LARVAL SURVIVAL AND GROWTH TEST
PERFORMED IN FORTY FATHOMS* ARTIFICIAL SEAWATER  USING
LARVAE FROM FISH MAINTAINED AND SPAWNED IN  FORTY  FATHOMS*
ARTIFICIAL SEAWATER,  AND SODIUM DODECYL SULFATE (SOS) AS A
REFERENCE TOXICANT',2,3,4,5
  Test
                 Survival
  NOEC
 (mg/LJ
                                            Growth
 NOEC
(mg/L)
 LOEC
(mg/L)
   Most
Sensitive
End Point
1
2
3
4
5
6
^Tests pe
1.0
1.0 -
1.0
1.0
1.0
1.0
rformed bv
•• • -- M MI -.- ,, ,
1.9
1.9
1.9
1.9
1.9
1.9
Donald .1 in
— - — i—
1.0
1.0
1.0
0.5
1.0
0.5

•
SE
SE
SE
1.0
SE
1.0
TV 	
G
G

                                            and
                   0997
3Adults collected in the field.
                         '=2
5For a discussion of the  precision  of  data from chronic toxicitv
 tests see Section 4,  Quality Assurance.        <-nromc toxicity
                                  78

-------
  TABLE  22.  SINGLE  LABORATORY  PRECISION  OF  THE  SHEEPSHEAD  MINNOW

            (CYPRINIDON  VARIEGATUS)  LARVAL  SURVIVAL  AND  GROWTH  TEST

            PERFORMED  IN NATURAL SEAWATER,  USING LARVAE  FROM  FISH

            MAINTAINED AND SPAWNED IN NATURAL SEAWATER,  AND COPPER AS A
            REFERENCE TOXICANT! 2 3  4 5   -,   ,
                               f f t f    --.:-
Survival
Test

1
2
3
4
5
NOEC
(ug/L)

125
125
125
125
250
LOEC
(ug/L)

250
250
250
250
500
Growth
NOEC
(ug/L)

125
31
125
125
125
LOEC
(ug/L)

SE
63
SE
SE
250
Most
Sensitive
End Point

G


G
 ]Tests  performed  by  George Morrison  and  Elise  Torello,  Environmental
 Research Laboratory, U. S. Environmental  Protection Agency,
 Narragansett, Rhode  Island.

2Three  replicate  exposure chambers,  each with  10-15 larvae, were
 used for the control and each copper concentration.  Copper
 concentrations were: 31, 63, 125, 250, and 500 ug/L.
3Adults collected in the field.

4SE = Survival effects.   Growth data at these toxicant concentrations
 were disregarded because there was a significant reduction in
 survival.

5For a discussion of the precision of data from chronic toxicity
 tests see Section 4, Quality Assurance.
                                  79

-------
 TABLE 23. SINGLE LABORATORY PRECISION OF THE SHEEPSHEAD MINNOW
           (PPRINIDON VARIEGATUS) LARVAL SURVIVAL AND GROWTH TEST
           PERFORMED IN NATURAL SEAWATER,  USING  LARVAE FROM  FISH

           =™ ™ ---«"  SEAWATER,  ^sSSL
                                AS A  REFERENCE
Test
  1
  2
  3
  4
  5
               Survival
              NOEC
             (mg/Lj
 LOEC
(mg/L)
     growth
           LOEC
           (mg/L)
               Most
            Sensitive
            End Point
             2.5
             2.5
             1.3
             2.5
             1.3
 5.0
 5.0
 2.5
 5.0
 2.5
  5
1.3
2.5
1.3
1.3
 SE
2.5
 SE
2.5
 SE
G
S
G
Narragansett, Rhode Island.          "tal  Protectio" A9*ncy,
 were: 0.3, 0.6  us     .nd 50
3Adults collected in the field.
                                                 S°S
                                         "-
                                 80

-------
      TABLE 24.  DATA FROM AN INTERL.ABORATORY STUDY OF THE
                 SHEEPSHEAD MINNOW LARVAL SURVIVAL AND
                 GROWTH TEST, USING AN INDUSTRIAL EFFLUENT
                 AS A REFERENCE TOXICANT!.2,3
I
i Test
; Laboratory A
; Test 1

Test 2

Laboratory B V
Test 1

Test 2

Laboratory C
Test 1

Laboratory D
Test 1

Test 2
•m
NOEC
{% Effluent)

3.2
3.2
3.2
3.2

3.2
3.2
3,2 o-::
3.2 .0

3.2
1.0

3.2
3.2
1.0
3.2 .
End
Point

G
S
G
S

G
S
G
S

G
S

G
S
G
S
     points: G = growth; S = survival.
Affluent concentrations were:  0.32, 1.0, 3 2. 10 0
 and 32.0%.                                         '
3From Schimmel,  1987.
                                 81

-------








-*->
i/)
-p
o
CD
tr vi
 -t-
> c
5- 0
Z5 U
i 	 I/!
rO CD
^ 1 ^
S-
fO
-7; «
^? £
- > c
—13 U^
OJ
I/) 13
a. >
O) S-

- ° ^ i < t- B Q o I/! || 4 r- Si 1 z o h- X UJ O 1 1 i j i f ! : " -LIVE LARVAE 1 ! .. IE ! | |- 0 ! c £ a in < ^ f i | 1 • ^ ir 11 < a 3f- < >- Q > " Mf AN LARV* InlAflVAE


-------
-o
 •  •»-
t-   ai
                                                             -j
                                                             a.
                                                                 O
                                                                 CL
                                                                                      S £
                                                                                       > ^
                                                                               z _
                                                                                             O
                                                                                             CJ
                                                                                                                                         UJ
                                                                                                                                         o
                                                                               83

-------
Test Dates:
Figure 8. Data forms for sheepshead minnow larval survival
          and growth test.  Dry weights of larvae.1

                          Species:
        Pan
         #
    Cone.
      &
     Rep
Initial
 Wt,
(mg)
Final
 Wt.
(mg)
Diff.
  #
Larvae
Av. Wt./
Larvae
 (mg)
     ^Adapted from:  Hughes,  Heber,  Schimmel,  and Berry,  1987.

                                          84

-------
Test Dates:
Figure 9. Data forms for  sheepshead  minnow  larval  survival
       and  growth  test.   Summary of test results. '

                       Species:                         	
Effluent Tested:
TREATMENT
# LIVE
LARVAE
SURVIVAL
CO/ i
( '«)
MEAN DRY WT
LARVAE (mg)
±S D
SIGNIF. DIFF,
FROM CONTROL
(o)
MEAN
TEMPERATURE
(OC)
-±S D
ME.AN SALINITY
000
±SD
AV. DISSOLVED
OXYGEN
(mg. L) ±S D
















































COMMENTS:
          ^Adapted from: Hughes, Heber,  S.  C.  Schimmel,  and Berry,  1987.

                                         85

-------
                                     SECTION 12

                                   TEST METHOD 1.2

                     SHEEPSHEAD MINNOW  (CYPRINODON VARIEGATUS)
                   EMBRYO-LARVAL SURVIVAL  AND  TERATOGENICITY  TEST
                                    METHOD TOPS

  1.   SCOPE  AND  APPLICATION

  1.1   This  method  estimates the chronic toxicity of effluents and receiuinn
  waters to  the  sheepshead minnow  (Cy^rinodon  vaMegatus "us In  embryo  and
  larvae™  a nine-day, static reneaTteltT Wlfes inc ude the
  synergistic, antagonists, and additive effects of all the chemical  physical
  and  biological components which adversely affect the physiological and       '
  biochemical functions of the test organisms.  The test is useful   n screenina
  ?£ tffh96"5 be°Te °r9an1sms are exposed during embryonic development   9
  The test has several advantages over the larval growth test because feedina i,
  not required and the larvae are not dried and weighed.                    9  S
 1.3  Single or multiple excursions in toxicity may not  be detected usina  ?4-h
 composite _samPles.._  Also,  because of the long sample  collection  period  9

  ?nh° rumint-?mp0^tS.S;7P^n9'  and  because the ^st  chambers  are  no? sealed
 SetecteS In the ?est  ' * degradable tox1cants in the  *>urce  may  not be     '


 1.4  This method should be restricted to use  by, or under the  supervision of
 professionals  experienced  in  aquatic toxicity testing.         >«pervision or,

 2 .   SUMMARY OF METHOD                              *v':.?%,

 2.1   Sheepshead minnow  embryos and larvae are  exposed in  a static  renewal
 system   from shortly after fertilization of the eggs through four  days
 posthatch (total  of  nine days), to different  concentrations of effluent or to
 receiving water.  Test  results'are based on the total  frequency of bo?h
 mortality and  gross  morphological deformities  (terata).

 3.  DEFINITIONS
    (Reserved for addition of terms at a later date).
     format used for this method was taken from Kopp,  1983.
 This method was adapted from materials provided by Terry Hollister  USEPA
Region 6 Laboratory, Houston, Texas, and from Birge and Black   1981!  Horninq
and Weber, 1985, and Hughes,  Heber,  Schimmel, and Berry,  198?!              9

-------
 4.  INTERFERENCES

 4.1  Toxic substances may be introduced by contaminants in dilution water,
 glassware, sample hardware,  and testing equipment (see Section 5,  Facilities
 and Equipment).

 4.2  Adverse effects of low  dissolved oxygen concentrations (DO),  high
 concentrations of suspended  and/or dissolved solids,  and extremes  of pH may
 mask the effect  of toxic substances.

 4.3  Improper effluent sampling and handling may adversely affect  test
 results  (see Section 8,  Effluent and  Receiving  Water  Sampling  and  Sample
 Handling).   ,

 4.4  Pathogenic  and/or predatory organisms in the dilution water and
 effluent may affect  test organism survival,  and confound test  results.

 5.   SAFETY

 5.1   See Section  3,  Health and  Safety.

 6-   APPARATUS AND EQUIPMENT                   „.,,,,

 6.1   Facilities for  holding  and  acclimating  test  organisms.

 6.2   Sheepshead minnow culture  unit —  see 7.13.1  below.   To perform
 toxicity tests on-site or  in the  laboratory,  sufficient  numbers of  newly
 fertilized eggs must  be  available,  preferably from an  inhouse  sheepshead
 minnow culture unit.   If necessary, embryos  can be obtained from outside
 sources  if shipped in  well oxygenated water  in  insulated containers.

 6.2.1  A test using  15 embryos  per  test vessel  and four replicates  per
 concentration, will require 360 newly-fertilized embryos at the start of the
 test  (Table  5).   A test with a minimum of  10  embryos per test  vessel and   "
 three replicates  per concentration, and with  five  effluent concentrations
 and  a control, will require a minimum of 180  embryos at the start of the
 test.

 6.3  Brine shrimp  (Artemia) culture unit — for feeding sheepshead minnow
 larvae in  the continuous culture unit (see 7.12 below).

 6.4  Samplers —  automatic sampler, preferably with sample cooling
 capability, that can collect a 24-h composite sample of 5 L, and maintain
 sample temperature at 4°C.

6.5  Environmental chamber or equivalent facility with temperature control
 (25+20C).

6.6  Water purification system — Millipore Super-Q, deionized water (DI) or
equivalent.

6.7  Balance  -- analytical, capable of accurately weighing to 0.0001 g.

                                      87

-------
6.8  Reference weights, Class S -- for checking the performance of the
balance.  The reference weights should bracket the expected weights of
reagents, and the expected weights of the weighing boats and the weights of
the weighing boats plus larvae, used in Artemia suitability studies.

6.9  Air pump — for oil free air supply.

6.10  Air lines, and air stones -- for aerating water containing embryos,
larvae, or supplying air to test solution with low DO.

6.11  pH and DO meters — for routine physical and chemical measurements.
Unless the test is being conducted to specifically measure the effect of one
of the above parameters, a portable, field-grade instrument is acceptable.

6.12  Standard or micro-Winkler apparatus -- for determining DO (optional).

6.13  Dissecting microscope — for examining embryos and larvae.

6.14  Light box — for counting and observing embryos and larvae.

6.15  Refractometer -- for determining salinity.

6.16  Thermometers, glass or electronic, laboratory grade — for measuring
water temperatures.

6.17  Thermometers, bulb-therrnograph or electronic-chart type -- for
continuously recording temperature.

6.18  Thermometer, National Bureau of Standards Certified (see USEPA METHOD
170.1, USEPA, 1979} -- to calibrate laboratory thermometers.

6.19  Test chambers —  four (minimum of three),  borosilicate glass or
non-toxic plastic labware per test concentration.   The chambers should be
covered during the test to avoid potential contamination from the air.  Care
must be taken to avoid inadvertently removing embryos or larvae when test
solutions are decanted from the chambers.  The covers are removed only for
observation and removal of dead organisms.

6.20  Beakers — six Class A, borosilicate glass or non-toxic plasticware,
1000 ml for making test solutions.

6.21  Wash bottles — for deionized water, for washing embryos from
substrates and containers, and for rinsing small  glassware and instrument
electrodes and probes.

6.22  Volumetric flasks and graduated cylinders — Class A, borosilicate
glass or non-toxic plastic labware, 10-1000 mL for making test solutions.

6,23  Pipets, volumetric — Class A, 1-100 mL.

6.24  Pipets, automatic — adjustable,  1-100 mL.
                                      88

-------
  6.25   Pipets,  serological  —  1-10 mL,  graduated.

  6.26   Pipet  bulbs and  fillers  — PROPIPET*, or equivalent.

  6.27   Droppers and glass tubing with fire polished aperatures, 4 mm ID
  for transferring embryos and  larvae.

  6.28   Siphon with bulb and clamp — for cleaning test chambers.

  6.29  ^NITEX* mesh sieves, < 150 urn, 500 urn, and 3-5 mm - for collecting
         and fish embryos, and for spawning baskets, respectively

                                           ' I8ubei
 7.  REAGENTS AND CONSUMABLE MATERIALS

 7.1  Sample containers — for sample shipment and storage (see Section 8
 Effluent and Receiving Water Sampling and Sample Handling).

 7.2  Data sheets (one set per test)  — for data recording (see Fig.  5).

 7.3  Tape,  colored ~ for labelling  test chambers

 7.4  Markers,  water-proof — for marking containers,  etc.

 7.5  Buffers,  pH 4,  7,  and 10 (or as per instructions  of  instrument
               f°r standards  and  calibration  check (see USEPA  Method  150.1,
 7.6   Membranes  and  filling  solutions  for dissolved oxygen  probe  (see  USEPA
 Method  360.1, USEPA,  1979), or  reagents for modified Winkler  analysis.

 7.7.  Laboratory quality assurance samples and standards for  the above
 methods.

 7.8   Reference  toxicant solutions (see Section 4, Quality Assurance).

 7.9   Reagent water  — defined as distilled or deionized water that does not
 contain substances  which are toxic to the test organisms (see paragraph 6 6
 above).                                                              r

 7.10  Effluent, surface water, and dilution water — see Section 7, Dilution
 Water, and Section  8, Effluent and Surface Water Sampling and Sample
 Handling.                                                        ^

 7.11  Saline test and dilution water -- The overwhelming majority of
 industrial and sewage treatment effluents entering marine and estuarine
 systems contain little or no measurable salts.  Exposure of sheepshead
minnow embryos to these effluents will require adjustments in the salinity
of the test solutions.  This test has been successfully performed over a
range of salinity of 6 °/00 to 59 °/oo salinity.   It is important to
                                      89

-------
 maintain a constant salinity across all  treatments.   Also, the salinity
 should vary by no more than + 2 °/oo among the chambers on a given day.

 7.11.1  If effluent and receiving water  tests are conducted concurrently,
 the salinities of these tests should be  similar.   In addition, it may be
 desirable to match the test salinity with that of the receiving water.  Two
 methods are available to adjust salinities — a supersaline brine derived
 from natural  seawater or artificial'sea  salts.

 7.11.2  Hypersaline brine (100 °/oo  salinity):   Hypersaline brine (HSB)
 has several  advantages that make it  desirable for use in toxicity testing.
 It  can be made from any high quality,  filtered  seawater by evaporation, and
 can be added  to the effluent or to deionized  water to increase the
 salinity.   HSB derived from natural  seawater  contains the necessary trace
 metals,  biogenic  colloids,  and some  of the microbial  components necessary
 for adequate  growth,  survival,  and/or  reproduction of marine and estuarine
 organisms,  and may be stored for prolonged periods without any apparent
 degradation.   However,  the  concentration  of effluent  that can  be tested
 using  HSB is  limited  to 80% at  20  o/oo salinity,  and  70% at 30% salinity.

 7.11.2.1   The  ideal  container  for  making  brine  from  natural  seawater  is one
 that  (1)  has  a high  surface to  volume  ratio,  (2)  is made of a  non-corrosive
 material,  and  (3)  is  easily cleaned  (fiberglass containers are ideal).
 Special  care  should  be  used to  prevent any toxic  materials from coming  in
 contact  with the  seawater being  used to generate  the  brine.   If a  heater is
 immersed  directly into  the  seawater, ensure that  the  heater  materials  do not
 corrode  or  leach  any  substances  that would contaminate  the brine.   One
 successful method  used  is a thermostatically  controlled  heat exchanger  made
 from fiberglass.   If  aeration  is used, use only oil-free air compressors to
 prevent  contamination.

 7.11.2.2   Before  adding  seawater to the brine generator,  thoroughly clean
 the generator,  aeration  supply tube, heater,  and  any  other materials that
 will be  in direct  contact with the brine.   A  good  quality  biodegradable
 detergent  should be used, followed by  several (at  least  three)  thorough
 deionized water rinses.

 7.11.2.3  High  quality  {and preferably high salinity) seawater  should be
 filtered to at  least  10  urn  before placing  into the brine generator.  Water
 should be collected on an incoming tide to minimize the  possibility of
 contamination.

 7.11.2.4  The  temperature of the seawater  is  increased slowly to 40°C.
The water should be aerated to prevent temperature stratification and to
 increase water evaporation.  The brine should be checked daily  (depending on
volume being generated)  to ensure that salinity does  not exceed  100 o/oo
and that the temperature does not exceed 40°C.  Additional seawater may be
added to the brine to obtain the volume of brine required.

7.11.2.5  After the required salinity  is attained, the brine should be
filtered a second time through a 1-um filter and poured directly into
portable containers, such as 20-1 (5-gal) cubitainers or polycarbonate water

-------
cooler jugs.  The containers should be capped and labelled with the date the
brine was generated and its salinity.  Containers of brine should be stored
in the dark and maintained at room temperature until used.

7.11.2.6  If a source of hypersaline brine is available, test solutions can
be made by following the directions below.  Thoroughly mix together the
deionized water and brine before mixing in the effluent.

7..11.2.7  Divide the salinity of the hypersaline brine by the expected test
salinity to determine the proportion of deionized water to brine.  For
example, if the salinity of the brine is 100 o/oo and the test is to be
conducted at 20 o/oo, 100 o/oo divided by 20 o/oo - 5.0.  The
proportion of brine Is 1 part in 5 (one part brine to four parts deionized
water).

7.11.2.8  To make 1 L of sea water at 20 o/0o salinity from a hypersaline
brine of 100 o/00, divide 1 L (1000 mL) by 5.0.  The result, 200 mL, is
the quantity of brine needed to make 1 L of sea water.  The difference, 800
ml, is the quantity of deionized water required.

7.11.2.9  Table 1 illustrates the composition of test solutions at 20 °/oo
if they are prepared by serial dilution of effluent with 20 °/oo salinity
seawater.

7.11.3  Artificial sea salts:  HW MARINEMIX& brand sea salts (Hawaiian
Marine Imports Inc., P.O. Box 218687, Houston, Texas 77218) have been used
successfully at the USEPA Houston laboratory to culture sheepshead minnows
and perform the embryo-larval survival and teratogenicity test.
EMSL-Cincinnati has found FORTY FATHOMSR artifical sea salts (Marine
Enterprises, Inc., 8755 Mylander Lane, Baltimore, Maryland 21204; phone:
301-321-1189), to be suitable for culturing sheepshead minnows and for
performing the larval survival and growth test and embryo-larval test.
Artificial sea salts may be used for culturing sheepshead minnows and for
the embryo larval test if the criteria for acceptability of test data are
satisfied (see Paragraph 12).

7.11.3.1  Synthetic sea salts are packaged in plastic bags and mixed with
deionized water or equivalent.  The important thing is to follow the
instructions on the package of sea salts carefully and to mix the salts in a
separate container -- not the culture tank.  The deionized water used in
hydration should be in the temperature range of 21-260C.  Seawater made
from artificial sea salts is conditioned (see Spotte, 1973; Bower, 1983)
before it is used for culturing or testing.  After adding the water, place
an airstone in the container, cover, and aerate the solution mildly for at
least 24 h before use.
                                      91

-------
TABLE 1.  PREPARATION OF TEST SOLUTIONS AT A SALINITY
          OF 20 o/oo USING 20 o/00 NATURAL OR ARTIFICIAL
          SEAWATER, HYPERSALINE BRINE, OR ARTIFICIAL SEA SALTS
                                           Solutions To Be Combined
Effluent
Effluent Cone.
Solution (%)
1 1001'2
2 32
3 10
4 3.2
5 1.0
Control 0.0
Total
Volume of Volume of Diluent
Effluent Seawater (20 o/00)
Solution
4000 mL ...
1000 mL Solution 1 + 2000 mL
1000 mL Solution 2 + 2000 mL
1000 mL Solution 3 + 2000 mL
1000 mL Solution 4 + 2000 ml
2000 ml
10000 mL
  'This illustration assumes: (1) the use of 400 mL of test
  solution in each of four replicates  (total of 1600 mL) for the control
  and five concentrations of effluent, (2) an effluent dilution factor of
  0.3, and (3) the effluent lacks appreciable salinity.  A sufficient
  initial volume (4000 mL) of effluent is prepared by adjusting the
  salinity to the desired level.  In this example, the salinity is adjusted
  by adding artificial sea salts to the 100% effluent, and preparing a
  serial  dilution using 20 o/00 seawater (natural  seawater, hypersaline
  brine,  or artificial seawater).  The salinity of the initial  4000 mL of
  100% effluent is adjusted to 20 o/00 by adding 80 g of dry artificial
  sea salts {HW MARINEMIX or FORTY FATHOMS^),  and  mixing for 1  h.   Test
  concentrations are then made by mixing appropriate volumes of salinity
  adjusted effluent and 20 °/oo salinity dilution  water to provide
  3000 mL of solution for each concentration.   If  hypersaline brine alone
  (100 o/oo)  is used to adjust the salinity of the effluent,  the highest
  concentration of effluent that could be achieved would be 80% at
  20 °/oo salinity,  and 70% at 30 °/oo salinity.

 2The same procedures would be followed in preparing test
  concentrations at other salinities between 20 °/oo and 30 °/oo:   (1)
  The salinity of the bulk (initial) effluent  sample would be adjusted  to
  the appropriate salinity using artificial  sea salts or hypersaline brine,
  and (2)  the  remaining effluent concentrations would be prepared  by serial
  dilution,  using a  large batch  (10  L)  of seawater for dilution water,
  which had been prepared at the same salinity as  the effluent, using
  natural  seawater,  hypersaline  and  deionized  water.

-------
  7.12  BRINE  SHRIMP  (ARTEMIA)  CULTURE  (see Peltier and Weber,  1985).

  1'12'J ^  lf a sheePshead continuous culture unit  is established, newly-
  hatched  Artenna nauplii will  be needed for feeding the larvae, and a brine
  ,Sn^rLCU]tKre un1* Sh0uld be PrePared.  Although there are many commercial
  sources  of brine shnmp cysts, the Brazilian or Colombian strains are
  currently preferred because the supplies examined have had low
  concentrations of chemical residues and produce nauplii  of suitably small
  size.  (One source that has been found to be acceptable  is Aquarium
  Products, 180L Penrod Ct., Glen Burnie, Maryland 21061,  phone 800-368-2507).

  7.12.2  Each  new batch of Artemia  cysts must be evaluated  for size
  (Vanhaecke and Sorgeloos,  1980,  and Vanhaecke et al.,  1980)  and  nutritional
  suitability (see Leger et  al., 1985,  1986)  against known suitable referent
  cysts by performing  a  side by side larval growth  test  using  the  "new" and
  reference" cysts.   The "reference"  cysts used  in the  suitability test may
  be  a  previous y tested and acceptable  batch of  cysts,  or may  be obtained
  from  the  Quality Assurance Branch,  Environmental Monitoring and Support
  f±rl ny>   H^T*]' °hl'°-  A Sample of "^-hatched Artemia nauplii
  from  each new batch  of cysts  should be chemically analyzedT-rhi Artemia
  o£»S^ n" K  n°/     ?Sed-1f the concentration of total organic chTo7n?T
  exceeds 0.15  ug/g wet  weight or the total concentration of organochlorine
  sef USCEP" 1982S.     """^ ^ U9/9 ^ "^  For ^alyt^cal methods
 7.12.3  Artemia nauplii are obtained as follows:
         Add 1 L of seawater, or a solution prepared by adding 35.0 a
         umodized salt (Nad) or artificial  sea salts to 1  L  of deionized
         water, to a 2-L separatory funnel, or equivalent.
         Add 10 ml Artemia_cysts to the separatory funnel  and  aerate for  24  h
         at 27 oc.   (Hatching time varies  with incubation  temperature and
         the geographic strain of Artemia  used.   See Peltier and Weber 1985
                                           for details on ArWa culture and
         After424  h,  cut  off  the  air  supply  in the separatory funnel.
         Artemia naupln  are  phototactic, and will concentrate at the bottom
         if  a dark  cloth  or paper towel  is placed over the top of the
         separatory funnel for 5-10 min.  To prevent mortality, do not leave
         the concentrated nauplii at  the bottom of the funnel more than 10
         mm without aeration.
    4.   Drain the  nauplii into a funnel fitted with a   150 urn Nitex screen,
         and rinse  with seawater or equivalent before u"se.

7.12.4  Testing Artemia nauplii as food for sheepshead minnow culturing.

7.12.4 1  The primary criterion for acceptability of each new supply of
brine shrimp cysts is the ability of the nauplii  to  support  good  survival
and growth of the  sheepshead  minnow larvae.   The  larvae  used to evaluate  the

-------
  suitability of the brine shrimp nauplii must be of the same geographical
  origin,  species,  and stage of development as those used routinely in the
  larval  survival  and growth tests.   Sufficient data to detect differences in
  survival  and growth should be obtained by using three replicate test
  vessels,  each containing a minimum of 15 larvae,  for each  type  of food.

  7.12.4.2   The feeding  rate and frequency,  test  vessels,  volume  of control

  SS^hWh" ?! ^ teSt'  and ?9e  °f the  naupHi  at  the  star? o?  She
  test, should be the same as used for  the routine  toxicity tests,

  t.'n2;4:3^  Re!ultS  °f,the bi:1ne Shr1mp  nutrition assay, where  there are only
  two treatments, can  be  evaluated statistically by  use of a  t-test.   The
  "new" food  is  acceptable if there  are  no statistically significant
  differences  in the  survival and growth  of  the larvae fed the  two  sources of
  naup iii.
 ™Hm               seven-da* survival of larvae should be 80% or greater,
 and (2) the average dry weight of larvae should be 0.60 mg or greater  if
 dried and weighed immediately after the test,  or (3) the average dry we ght
 of larvae should be 0. 50 mg or greater, if the larvae are preserved in 4%
 formalin before drying and weighing.  The above minimum weights presume that
 the age of the larvae at the start of the test is not greater than 24 h.

 7.13   SHEEPSHEAD MINNOWS

 7.13.1   Brood Stock

 7.13.1.1   Adult sheepshead minnows for  use  as  brood  stock  may be obtained  by
 seine  in  Gulf of Mexico  and Atlantic coast  estuaries,  from commercial
 sources,  or  from young fish raised to maturity in  the  laboratory.   Feral
 inb?eedin     *""      9ene™tion  laboratory fish  are  preferred,  to minimize


 7.13.1.2   To  detect  disease and to allow  time  for  acute mortality  due  to the
 stress of  capture, field-caught adults  are  observed  in the  laboratory  a
 minimum of two  weeks before using  as  a  source  of gametes.   Injured  or
 diseased fish are discarded.                                  J

 7.13.1.3   Sheepshead minnows can be  continuously cultured  in  the laboratory
 from eggs  to adults.   The  larvae,  juvenile, and adult fish should be kept  In-
 appropriate size rearing tanks, maintained  at  ambient laboratory
 temperature.  The larvae should be fed  sufficient newly-hatched Artemia
 nauplii daily to assure that live  nauplii are  always present.  At the
 juvenile stage  they are fed frozen adult brine shrimp and a commercial
 flake food, such as TETRA SH-SO*,   available from Tetra Sales  (U.S A)  201

IrawEiwft r°rriS i^'ri6? JerS6y °7950' phone 800-526-0650, o/MARDEL
AQUARIAN* Tropical Fish Flakes, available from Mardel Laboratories, Inc.,
 1958 Brandon Court,  Glendale. Heights, Illinois  60139, phone 312-351-0606  or
equivalent.  Adult fish are fed flake food three or foSr times da? y
supplemented with frozen  adult brine shrimp.
                                      94

-------
 7.13.1.3.1  Sheepshead minnows reach sexual maturity in three-to-five months
 after hatch,  and have an average standard length of approximately 27 mm for
 females and 34 mm for males.   At this time, the males begin to exhibit
 sexual  dimorphism and initiate territorial behavior.  When the fish reach
 sexual  maturity and are to be used for natural  spawning,  the temperature
 should  be controlled at 18-2Q°C.

 7.13.1.4  Adults can be maintained in natural  or artificial sea water In an
 a  flow-through or recirculating, aerated system consisting of an all-glass
 aquarium, or  a "Living Stream" (Figid Unit, Inc.,  3214 Sylvania Ave,  Toledo,
 Ohio  43613, phone 419-474-6971), or  equivalent.

 7.13.1.5  The system is equipped with an undergravel or outside biological
 filter  of shells (see Spotte,  1973 or bower,  1983  for conditioning  the
 biological  filter),  or a cartridge filter,  such  as  a MAGNUM^ Filter,
 available from Carolina Biological Supply Co.,  Burlington,  North Carolina
 27215,  phone  800-334-5551,  or  an EKEIM*  Filter,  available  from Hawaiian
 Marine  Imports Inc.,  P.O.  Box  218687,  Houston,  Texas 77218,  phone
 713-492-7864,  or equivalent, at  a  salinity of 20-'30  
-------
 7.13.2.3.2  Each female is injected intraperitoneally with 50 IU HCG on two
 consecutive days, starting at least 4 days prior to the beginning of a
 test.  Two days following the second injection, eggs are stripped from the
 females and mixed with sperm derived from excised macerated testes.   At
 least ten females and five males are used per test to ensure that there is a
 sufficient number of viable embryos.

 7.13.2.3.3  HCG is injected into the peritoneal cavity, just below the skin,
 using as small  a needle as possible.  A 50 IU dose is recommended for
 females approximately 27 mm in standard length.  A larger or smaller dose
 may be used for fish which are significantly larger or smaller than  27 mm.
 With injections made on days one ,and two,  females which are held at  25°C
 should be ready for  stripping on Day 4.  Ripe females should show pronounced
 abdominal swelling,  and release at  least a few eggs in response to a gentle
 squeeze.   Injected females should be isolated from males.   It may be helpful
 if  fish that  are to  be injected are maintained at 20°C before injection,
 and the temperature  raised to 25<>C  on the  day of the first injection.

 7.13.2.3.4 Prepare  the testes immediately before stripping the eggs from
 the females.  Remove the testes from three-to-five males.   The testes  are
 paired,  dark  grey organs along the  dorsal  midline of the  abdominal cavity.
 If  the head of  the male is cut off  and  pulled away from the rest  of  the
 fish,  most of the internal  organs can  be pulled out of the body cavity,
 leaving  the testes behind.   The testes  are placed in a few ml  of  seawater
 until  the  eggs  are ready.

 7.13.2.3.5 Strip the  eggs  from the  females,  into a dish  containing  50-100  '
 ml  of  seawater,  by firmly  squeezing  the abdomen.   Sacrifice  the females  and
 remove the ovaries if  all  the  ripe  eggs do not  flow out freely.   Break up
 any clumps  of ripe eggs  and  remove  clumps  of  ovarian  tissue  and underripe
 eggs.   Ripe eggs  are  spherical,  approximately 1  mm in  diameter, and  almost
 clear.

 7.13.2.3.6  While  being  held over the dish  containing  the  eggs, the  testes
 are  macerated in  a fold  of  NITEX^ screen (250-500  urn mesh).dampened  with
 seawater.   The testes  are then  rinsed with  seawater  to  remove  the  sperm from
 tissue, and the  remaining  sperm  and  testes  are  washed  into  the dish.  Let
 the  eggs and milt  stand  together for  10-15  min,  swirling occasionally.

 7.13.2.3.7  Pour  the contents  of the dish  into  a  beaker, and  insert  an
 airstone.  Aerate gently, such that  the water moves  slowly over the  eggs,
 and  incubate at 250C for 60-90 min.   After  incubation, wash the eggs on a
 Nitex  screen and resuspend them  in clean seawater.

 7.13.2.4  Natural Spawning

 7.13.2.4.1  Short-term  (Demand) Embryo  Production

 7.13.2.4.1.1  Adult fish should be maintained at  18-2QOC in a temperature
controlled system.  To obtain embryos for a test, adult fish (generally, at
 least eight-to-ten females and three males) are transferred to a spawning
                                      96

-------
   chamber, with  a  photoperiod of  14 h  light/10 h dark and a temperature of
   250C, two days before the beginning  of the test.  The spawning chambers

   h^p?Pn?TfelyM??rvRX22 um M2h (HanSen' et al" ]978)> an< consist of a
   basket of 3-5 nw NITEX* mesh, made to fit into a 57-L (15 gal) aquarium.
   Spawning generally will begin within 24 h or less.  The embryos will fall
   through the bottom of the basket and onto a collecting screen (250-500 um
  Pmhrlrth^^10" ^ ""£*•  The Col1ect™9 tray should be checked for
  embryos the next morning.  The number of eggs produced is highly variable
  The number of spawning units required to provide the embryos needed to
  perform a toxicity test is determined by experience.  If the trays do not
  contain sufficient embryos after the  first 24 h,  discard the embryos!
  replace the  trays,  and collect the  embryos for  another 24  h  or less   To
  help keep the  embryos  clean,  the adults  are  fed  while  the  screens  are
  removed.


  fhJ3;f:?'K?h  The,?mbrros  fre  collected  in a  tray  placed on  the bottom of
  the tank   The  collecting  tray consists of <  150 um  NITEXR screen  attached
  to  a rigid plastic frame.  The collecting  tfays with newly-spawned, embryos
  are removed  from  the spawning tank, and the embryos  are collected  from the
  screens by washing them with a wash bottle or removing them with a fine
  brush.  The  embryos from several spawning units may  be pooled  in a single
  container to provide a sufficient number to conduct the test(s)   The
  embryos are  transferred into a petri dish or equivalent, filled'with fresh
  culture water, and are examined using a dissecting microscope or other
  suitab e magnifying device.  Damaged and infertile eggs are discarded (see
  Mg.  I).  It  is strongly recommended that the embryos be obtained from fish
 cultured inhouse,  rather than from outside sources, to  eliminate the
 uncertainty of damage caused  by shipping  and  handling that  may not  be
 observable,  but which might affect the results of the test.

 7.13.2.4.1.3   After  sufficient  embryos are collected  for  the  test,  the adult
 fish are  returned  to  the  (18-20<>C) culture tanks.

 7.13.2.4.2  Sustained Natural Embryo Production

 7.13.2.4.2.1  Sustained  (long-term), daily, embryo production can be

 Ll^NrlTRFAMR1^01"? ™ture f1sh 1n l™ks>  Such as a {285-L  Or 75-gal)
 LIVING STREAM* tank, at a temperature  of 23-25°C.  Embryos are  produced
 daily, and when needed, embryo "collectors" are placed on the bottom of the
 tank  on the afternoon preceding the start of the test.  The next morning,
 the embryo collectors are removed and the embryos are washed into a shallow
 glass culture dish using artificial seawater.

 7.13.2.4.2.2  Four embryo collectors,  approximately 20 cm X 45 cm,  will
 approximately cover the bottom of the 285-L tank.   The collectors are
fabricated from plastic  fluorescent light  fixture diffusors  (grids), with
cells approximately 14 mm deep X 14 mm square.   A screen consisting  of 500
urn mesh is  attached to one  side  (bottom)  of the grid with  silicone

                                    °f  the  9r1d pr°tects the

-------
  7.13.2.4.2.3   The  brood  stock  is replaced  annually with feral  stock.

  7.13.2.5  Test Organisms

  7.13.2.5.1  Embryos spawned over less than an 24-h period, are used for the
  test.  These embryos may be used immediately to start a test or may be
  placed in a suitable container and transported for use at a remote
  location.  When overnight transportation is required, embryos should be
  obtained when they are no more than 8-h old.  This permits the tests at the
  remote site to be started with less than 24-h old embryos.  Embryos should
  be transported or shipped in clean, insulated containers,  in well aerated or
  oxygenated fresh sea water or aged artificial sea water of correct salinity
  and should be protected from extremes of temperature and any other stressful
 conditions during transport.  Instantaneous changes  of water temperature
 when embryos are transferred from culture unit water to test dilution water
 or from transport container water to on-site test dilution,  should be less '
   5™  u  Jnstjntaneous changes  of PH,  dissolved ions,  osmotic strength,
 and DO should  also be  kept to a minimum.

 7.13.2.5.2  The number of embryos needed  to start the test will  depend  on
 the number of  tests to be conducted and  the objectives.  If  the test  is
 conducted with  four replicate test  chambers (minimum  of  three)  at  each
 toxicant  concentration and  in the control,  with  15 embryos {minimum of  10}
 in each  test chamber,  and the combined mortality  of embryos  prior  to  the
 start  of  the test  is less than  20%,  400 viable embryos are required for the
 t es v.

 8-  SAMPLE  COLLECTION,  PRESERVATION  AND HANDLING

 8.1  See  Section 8, Effluent  and Receiving  Water  Sampling  and Sample
 Handling.                                                        M

 9.  CALIBRATION AND STANDARDIZATION

 9.1  See Section 4, Quality Assurance.

 10.  QUALITY CONTROL

 10.1  See Section 4, Quality Assurance.

 11.  TEST PROCEDURES

 11.1  TEST SOLUTIONS

 11.1.1   Surface Waters

 11.1.1.1  Surface water toxicity is determined with samples used directly as
collected.
                                      98

-------
  11.1.2   Effluents

  11.1.2.1  The  selection of the effluent test concentration should be based
  on the objectives of the study.  One of two dilution factors, approximately
  0.3 or 0.5,  is commonly used.  A dilution factor of approximately 0.3 allows
  testing  between  100% and U effluent using only five effluent concentrations
  (100%, 30%,  10%, 3%, and 1%).  This series of dilutions minimizes the level
  of effort, but because of the wide interval between test concentrations
  provides poor test precision.  A dilution factor of 0.5 provides greater
  precision, but requires several additional dilutions to span the same range
  of effluent concentrations.  Improvements in precision decline rapidly as
  the dilution factor is increased beyond 0.5.

  11.1.2.2  If the effluent is known or suspected to be highly toxic,  a lower
 range of effluent concentrations should be used (such as 10%, 3%, 1%, 0.3%s
 and 0.1k).

 11.1.2.3  The volume of effluent required  to initiate the test and for  daily
 renewal  of four replicates  {minimum of three)  per concentration  for  five
 concentrations of effluent  and  a control,  each  containing 400 mL of  test
 solution, is  approximately  4  L.   Prepare enough  test  solution (approximately
 3000  ml)  at  each  effluent concentration to refill  the test  chambers  and
 provide  at least  400  ml additional  volume  for chemical  analyses.

 11.1.2.4   Maintain  the  effluent  at  4<>C.  Plastic  containers  such  as  8-20 L
 cubitainers have  proven successful  for  effluent collection  and storage.

 11.1.2.5   Approximately one hour  before  use, warm  a sufficient volume of
 chilled effluent  or receiving water sample(s) to the  test temperature
 (25 +  2°C) and  maintain it at that  temperature until portions are  added  to
 the dilution  water.

 11.1.2.6   The higher effluent concentrations {i.e., 10, 32, and  100%) may
 require aeration  to maintain adequate dissolved oxygen concentrations.
 However,  if one solution is aerated, all concentrations must be aerated.
 Aerate effluent as it warms and continue to gently aerate test solutions in
 the test  chambers for the duration of the test.

 11.2  START OF THE TEST

 11.2.1  Tests should begin as soon as possible,  preferably within 24 h after
 sample collection.  For on-site toxicity studies,  no more than 24 h should
 elapse between collection of the effluent and use in a embryo-larval  study.
 If the persistence of the sample toxicity is not known, the maximum holding
 time should not exceed 36 h  for off-site toxicity studies.  In no case
 should the test be started more than 72 h after  sample collection.

 11.2.2  Label  the test chambers with a marking  pen and identify each
treatment and  replicate with various colored coded tape.   A minimum of five
                                       99

-------
 effluent concentrations and a control should be selected for each study.
 Each concentration (including controls) is to have four replicates (minimum
 of three).  Use 500 ml beakers, crystallization dishes, non-toxic disposable.
 plastic labware, or equivalent for test chambers.

 11.2.3  Prepare the test solutions (see Table 1) and add to the test
 chambers.

 11.2.4  Gently agitate and mix the embryos to be used in the test in a large
 container  so that eggs from different spawns are evenly dispersed.

 11.2.5  The test is started by placing embryos from the common pool,  using a
 small  bore (2mm),  fire polished,  glass tube calibrated  to contain
 approximately the desired number  of embryos, into each  of four replicate
 test Chamber in sequential order,  until  each chamber contains 15 embryos
 (minimum of 10),  for  a total  of 60 embryos for each  treatment (four
 replicates recommended,  three minimum).   The amount  of  water added to  the
 chambers when transferring the embryos should be kept to a minimum to  avoid
 unnecessary dilution  of  the test  concentrations.

 11.2.6  After the  embryos have been distributed  to each  test chamber,
 examine  and count  them.   Remove and discard damaged  or  infertile eggs  and
 replace  with undamaged embryos.   It may  be more  convenient and  efficient to
 transfer embryos to intermediate containers of dilution  water for
 examination and counting.   After the  embryos  have been examined and counted
 in the intermediate container,  assign  them to the appropriate test chamber
 and  transfer them  with a  minimum of dilution  water.

 11.2.7  Randomize  the  position  of  the  test chambers  at the beginning of  the
 test   (see Appendix). Maintain the chambers  in this  configuration
 throughout  the  test.   A position chart may be helpful.

 11.3   LIGHT,  PHOTQPERIOD,  TEMPERATURE, AND  SALINITY

 11.3.1   The  light  quality  and  intensity  should be at  ambient  laboratory
 levels,  approximately  10-20 uE/iWs, or  50  to 100 foot candles  (ft-c),
with a photoperiod  of  14 h of  light and  10  h  of darkness.  The  test water
temperature  should  be  maintained at 25 + 2<>C.  The salinity  should be 5 to
32+2 o/oo  to  accommodate receiving waters that may fall within this
range.

11.4   DISSOLVED OXYGEN CONCENTRATION (DO)

11.4.1  Aeration may affect the toxicity of effluents and should be used
only as a last resort  to maintain satisfactory DO.  The  DO should not fall
below 60% saturation.   If  it is necessary to aerate, all treatments and the
control should be aerated.  The rate should not exceed 100 bubbles/min,
using a pipet with a 1-2 mm orifice, such as a 1-mL Kimax Serological Pipet
No. 37033,  or equivalent.  Care should be taken to ensure that turbulence
resulting from the aeration does not cause undue physical stress to the fish.
                                      100

-------
11.5  FEEDING                               .

11.5.1  Feeding is not required.

11.6  TEST SOLUTION RENEWAL

11.6.1  The test solutions are adjusted to the correct salinity and renewed
daily using freshly collected samples.  During the daily renewal process, 7-10
mm of water is left in the chamber to ensure that the embryos and larvae
remain submerged during the renewal process.  New test solution (400 mL)
should be added slowly by pouring down the side of the test chamber to avoid
exposing the embryos and larvae to excessive turbulence.

11.6.3  Prepare test solutions daily, making a minimum of five concentrations
and a control.  If concurrent effluent and receiving water testing occurs, the
effluent test salinity should closely approximate that of the receiving water
test.  If an effluent is tested alone, select a salinity which approximately
matches the salinity of the receiving waters.   Table 1 illustrates the
quantities of effluent, sea water, deionized water, and artificial sea salts
needed to prepare 3 L of test solution at each effluent concentration for
tests conducted at 20 o/oo salinity.

11.7  ROUTINE CHEMICAL AND PHYSICAL DETERMINATIONS

11.7.1  At a minimum, the following measurements are made and recorded  (see
Figure 5).

11.7.1.1  DO is measured at the beginning and end of each 24-h exposure period
at all test concentrations and  in  the control.

11.7.1.2  Temperature, pH, and  salinity are measured at the end of each 24-h
exposure period at all test concentrations and in the control.

11.8  OBSERVATIONS DURING THE TEST

11.8.1  At the end of the first 24 h  of exposure, before renewing the test
solutions, examine and count the  embryos.  Remove the dead embryos  (milky
colored and opaque) and record  the number.  If the  rate of mortality or fungal
infection exceeds  20%  in the control  chambers, or if excessive
non-concentration-related-mortality occurs, terminate the test  and  start  a  new
test  with new  embryos.   If the  above  mortality conditions do  not  occur,
continue  the  test  for  the  full  nine days.

11.8.2  At  25°C,  hatching  begins  on about  the sixth day.  After hatching
begins, count  the  number of  dead  and live  embryos and the number  of  hatched,
dead, live,  and  deformed and/or debilitated  larvae, daily  (see  Figure  1  for
illustrations  of  morphological  development of embryo and  larva).   Deformed
larvae  are  those  with  gross  morphological  abnormalities such  as curved
                                        101

-------
Figure 1. Embryonic developnent of sheepshead minnow (Cyprlnodon
variegatus):  A. Mature unfertilized egg, showing attachment filaments
and micropyle, 133; 3. Blastodisc fully developed;  C,D. Blastodisc,
8 cells; E. Blastoderin, 16 cells; F. Blastoderm,- late cleavage stage;
G. Blastoderm with germ ring formed,, embryonic shield developing;
H. Blastoderm covers over 3/4 surface of yolk, yolk noticeably
constricted;  I. Early embryo.  (From Kuntz, 1916.)
                                  102

-------
                                       '/JftP^g^g
Flaure 1
                                 0

           (Continued).   Embryonic development of sheepshead minnow
            variegatus):  J.  Embryo 48  h  after fertilization, now
                              on yolk  sac and body, otoliths formed
seo
K  Posterior portion of embryo  free  from yolk and moves freely, within
egg Sane, 72 h after fertilization; L.  Newly hatched fish, actual
llngth 4 mm; M. Larval  fish 5 days after hatching, actual  length 5 mm;
N? Young fish 9 mm in length; 0.  Young  fish 12 mm in  length.  (From
Kuntz, 1916.)                        :1=

                                  103

-------
 spines,  lack  of  appendages,  lack  of  fusiform  shape  (non-distinct  mass),  a
 colored  beating  heart  in  an  opaque mass,  lack of mobility,  abnormal
 swimming,  or  other  characteristics that preclude survival.   Remove dead
 embryos  and dead and deformed  larvae as previously  discussed and  record
 the  numbers for  all test  observations (see  Figure 5).

 11.8.3   Protect  the embryos  and larvae from unnecessary disturbance  during
 the  test by carefully  carrying out the daily  test observations, solution
 renewals,  and removal  of  dead organisms.  Make  sure the test organisms
 remain immersed  during the performance of the above operations*

 11.9  TERMINATION OF THE  TEST

 11.9.1   The test  is terminated after  nine days  of exposure,  or four
 post-hatch, whichever  comes  first.   Count the  number of.surviving,
 and deformed and/or debilitated larvae, and record the numbers of each.
 The deformed larvae are treated as dead.  Keep  a separate record of  the
 total number of  deformed  larvae for  use in  reporting the teratogenicity of
 the test solution.

 12.  ACCEPTABILITY OF  TEST RESULTS

 12.1  For  the test results to be acceptable,  survival in the controls must
 be at least 80% or better.

 13.  SUMMARY OF TEST CONDITIONS

 13.1  A  summary of test conditions is listed  in Table 2.

 14.  DATA ANALYSIS

 14.1  GENERAL

 14.1.1  Tabulate and summarize the data.

 14.1.2  The endpoints of this toxicity test are based on  total mortality,
combined number of dead embryos,  dead larvae,  and deformed larvae.  Point
estimates,  such as LCI, LC5,  LC10  and LC50,  are calculated using Probit
Analysis (Finney, 1971).   LOEC and NOEC values, for  total  mortality,  are
obtained using a hypothesis test  approach  such as Dunnett's Procedure
 (Dunnett,  1955)  or Steel's Many-one Rank Test  (Steel,  1959; Miller,
 1981).  See the Appendix for  examples of the manual  computations,  program
 listings, and examples  of data input  and program output.

14.1.3  The statistical tests described here must be used  with a knowledge
of the assumptions upon which the  tests are contingent.   The assistance of
a statistician is recommended for  analysts who are  not proficient  in
statistics.
                                     104

-------
    TABLE  2.   SUMMARY  OF RECOMMENDED  TEST CONDITIONS  FOR  THE  5HEEPSHEAD
              MINNOW  (CYPRINODON  VARIE6ATUS)  EMBRYO LARVAL  SURVIVAL  AND
              TERATOGENICITY TEST
   1.  Test type:
   2.  Salinity:
   3.  Temperature:
   4.  Light quality:
   5.  Light intensity:
   6.   Photoperiod:
   7.   Test  chamber  size:
   8.   Test  solution volume:
   9.   Renewal of test concentration
 10.  Age of test organisms:
 11.  No. of embryos/chamber:
 12.   Replicate test chambers/
      concentration:
 13.   Embryos  per concentration:
 14.   Feeding  regime:
 15.   Aeration:

 16.   Dilution water:


17.  Effluent test concentrations:
18.  Dilution factor:
19.  Test duration:
  Static renewal
  5 o/oo to 32 o/oo t 2
  25 + 20C
  Ambient laboratory light
  10-20 u£/m2/s,  or 50-100 ft-c
  {ambient  laboratory levels)
  14 h  light,  10  h  dark
  500 mL
  400 mL  (minimum of 250 mL)
  Daily
  less than 24 h old
  15 (minimum of 10)

 4 (minimum of 3)
 60  (minimum.of 30)
 Feeding  not  required
 None unless  DO falls below 60%
 saturation
 Uncontaminated source of  sea water;
 deionized water mixed with
 artificial sea salts, or  hypersaline
 brine          ;"::
 5 and a control
Approximately 0.3 or 0.5
9 days

-------
                            TABLE 2.  CONTINUED
20.  Effects measured;
Percent hatch; percent larvae dead
or with debilitating morphological
and/or behavior abnormalities such
as: gross deformities; curved spine;
disoriented, abnormal swimming
behavior; surviving normal larvae
from original embryos
                                    106

-------
   14.2
fEXRA?OGEN°K?TYLDYATA °F SHEEPSHEAD MINNOW EMBRYO-LARVAL SURVIVAL AND
   14.2.1  Formal statistical analysis of the total  mortality data  is outlined in
   Fiqure 2.  The response used in the analysis is the  total  mortal ty orooortion
   in each test or control chamber.   Separate analyses  are performed for ?he

            and iW'S "• ^r^ p01'nts ^ for the es'tiLtion o? the LCI
            and LC50 end points.  Concentrations at  which there is  10OT total

      h ' k™an VoEVhLtet,hHarrS  '?  6XCluded from **«1st,caTa a° g is
     the NUEC and LOEC,  but  included  in  the  estimation of the LC end points.
  14:2Zu  For th^ case  of  equa1 n^bers of replicates across all concentrations
  and the control,  the  evaluation of the NOEC and LOEC end points is rode  la  1
  parametric  test,  Dunnett's Procedure, or a nonparametric test  StLrf

  Many-one Rank  Test, on the arcsin transformed data   Underlying asslptions  of

  tested * SThP°5«tr?>'n0riBa1^ 8nd hom°9e"e-°"e Ra"k  Test,  is ued  to determine
  mp?  ?hS f H    • .6nd P°lnt-'  If the ass™Pti°"s  of Dunnett's  Procedure"
  met, the end points are estimated by the  parametric  procedure.   °ceaure are
 tPH             numbe:s.of ^Plicates  occur among the concentration levels
 tested, there are parametric and  nonparametric alternative analyses   The

 tPhpaSonflC anal^S JS '^.Bonferronl  t-test.  The Wilcoxon Ra k Sum Test with
 the Bonferrom adjustment  is the  nonparametric alternative.  For deta led
 information on the Bonferroni  adjustment  see the Appendix!       aetalled


 14.2.4  Probit Analysis  {Finney,  1971) is used to estimate the concentration
 that causes a  specified  percent decrease  in survival  from the control

              '                      ata ^ a11                      -
                                          STSS.S   .
number of embryos at start of test),  is the response of  interest!  The total
mortality proportion in each replicate must first be transformed by the arcsin
transformation procedure described  in  Appendix  C.  The raw and transformed
data  means and standard deviations of the  transformed observations aTeach
SDS concentration and control  are  listed  in Table 3.  A plot of the data is
provided in Figure 3.  Since there is  100%  total mortality in both replicates
for the 8.0 mg/L concentration,  it is  not included in this statistical
analysis and is considered  a qualitative mortality effect.
                                     107

-------
            TABLE 3.  SHEEPSHEAD MINNOW EMBRYO-LARVAL TOTAL MORTALITY DATA
           Replicate
   RAW
       Control

         0.1
         0.1
                                           SDS Concentration tmq/L
         0.5
         - • •  !•

         0.0
         0.2
1.0


0.0
0.1
2.0


0.3
0.1
4.0


0.9
0.7
8.0


1.0
1.0
ARC SINE
TRANS- A
FORMED B
~ 	 Mil 	
MEAN(Yt)
Si2
i
0.322
0.322
-^~ '" ~- • •• - ••
0.322
0.0
1
0.159
0.464
0.311
0.046
2
0.159
0.322
0.240
0.013
3
0.580
0.322
0.451
0.033
4
1.249
0.991
1.120
0.033
5
 14.2.6  Test for Normality
 Jh;2;f J  /inCe on]Vw9  rePlicat«  were run  at each  concentration level
 the  test  for  normal 7 ty is  invalid.   Additionally,  a non-parametric
 alternative to  Dunnett's  Procedure is  not  available with  only
 duplicates.   Thus  the  only information that can be derived  from the  data
 is from Dunnett's  Procedure.  However,  the results from !h is test  should
 quesS           CaUtl'°n SlnCe thS  ass"mPt1o'ls  of  the  test  are  ?n

 14.2.7  Dunnett's Procedure

14.2.7.1  To obtain an estimate of the pooled variance for the Dunnett's
Procedure, construct an ANOVA table as described in Table 4
    Source
                           TABLE 4.   ANOVA TABLE
df
Sum of Squares
     (SS)
Between
Within
— i — - • •
Total
ii ni..
p - 1
N - p
N - 1
'" ' 	 - • -
SSB
SSW
SST
                                                  Mean  Square (MS)
                                                      (SS/df)
                                                 % = SSB/(p-l)
                                                  2
                                                 SW - SSW/(N~p)
                                   108

-------
    SIATISTI"L
                             TOTAL MORTALITY
                      TOTAL NUMBER OF DEAD EMBRYOS
                     DEAD LARVAE AND DEFORMED LARVAE
                               ARCSIN
                           TRANSFORMATION
          ESTIMATE


             NORMAL DISTRIBUTION

                        HETEROGENEOUS
                           VARIANCE
                           BARTLETT'S
HOMOGENEOUS VARIANCE
              EQUAL NUMBER OF
                REPLICATES?
          EQUAL NUMBER OF
            REPLICATES?
T-TEST WITH
BONFERRONI
ADJUSTMENT
  WILCOXON RANK SUM
      TEST WITH
BONFERRONI ADJUSTMENT
                DUNNETT'S
                  TEST
STEEL'S MANY-ONE
   'RANK TEST
                         ENDPOINT ESTIMATES
                              NOEC. LOEC
 Figure 2.  Flow chart for statistical analysis of sheeoshead
          minnow embryo-larval data.               sneepsnead

                          109

-------
    S-
    >
   i.
   O
  (O
  -l-»
  O
 -a
  ro
 Q.
 CD
 CD
 O

CL,
ro

-------
  Where;
              n,-  = number of observations in concentration'!
            SSB  - i Tl2/nf  -  G2/N
                                         Between Sum of Squares
         SST = £    rYij2 . G2/N

         SSW = SST - SSB
                                        «i vii in OUHI UT squares
          G  = the grand total of all sample observations, G - I
                                           Total Sum of  Squares

                                           Within Sum of Squares
                                                          for
               = the jth observation for  concentration  "i"
14.2.7.2  For the data in this example:

    "1  = n2  = n3  = n4 = n$ = 2
    N   = 10
    1]  =  Y]]  +  Yi2  -f Y]3  =  0.644
   T2  =  Y2]  +  Y22  + y23  .  0>623
   |3  -  Y31  +  Y32  + Y33  =  0.481
   |4  ~  Y4i  +  Y42  + Y43  =  0.902
    "  -  Y51  +-Y52  + Y53  =  2.240
       = T] + T2 +  73 + T4  - 4.890
G

SSB
  P
= I
 1 = 1
                    G2/N
       = _1_(6.865)  -  (4.890)2  a ^  041'
          2               10
  SST =
     3.559 - (4.890)2  =
                10
                              168
                                  111

-------
       SSW = SST - SSB = 1.168 - 1.041 = 0.127
       SB 2 = SSB/p=l  = 1.041/5-1  - 0.260
       Sw 2 = SSW/N-p  = 0.127/10-5  = 0.025

   14.2.7.3  Summarize these  calculations in  the ANOVA  table  (Table 5)

             TABLE 5.   ANOVA TABLE  FOR DUNNETT'S  PROCEDURE EXAMPLE
      Source
          df
Sum of Squares
     (SS)
                                                    Mean Square(MS)
                                                        (SS/dfJ
           To perform the individual comparisons, calculate the t
           for each concentration, and control combination as follows

                                     (  YT - 7i )
 Where
      Sw

      nl
      n
= mean proportion of total mortality for SDS concentration i
- mean proportion of total mortality for the control
- square root of within mean sqaure
= number of replicates for control
= number of replicates for concentration i
mortal it arS 10°kin9 for an increase<* response in percent of
        j wvci  L.uFiLiuij me control me a n is c\tv\'¥v*r++>nft f^.^^.
                                                                     at a
14.2.7.5  Table 6 includes the calculated t values for  each
                            { 0.311  -  0.322  )
                       0.158VTT7FT+TT72T ]
                                                = 0.570
                                   112

-------
                         TABLE 6.  CALCULATED T-VALUES
             SDS Concentration Cmg/L)
0.5
1.0
2.0
4.0
2
3
4
5
-0.070
-0.519
0.816
5.051
 14.2.7.6  Since the purpose of this test is to detect a significant
 increase in total mortality, a {one-sided} test is appropriate.  The
 critical value for this one-sided test is found in Table 5, Appendix C
 For an overall alpha level of 0.05, five degrees of freedom for error and
 four concentrations (excluding the control) the critical value is 2.85.
 The mean proportion of total mortality for concentration "i" is
 considered significantly less than the mean proportion of total mortality
 for the control if tj  is greater than the critical value.   Therefore,
 only the 4.0 mg/L concentration has a significantly higher mean
 proportion  of total  mortality than the control.   Hence the NOEC is  2 0
 mg/L and the LOEC is 4.0 mg/L.

 14.2.7.7  To quantify  the sensitivity of  the  test,  the minimum
 significant  difference  (MSD)  that  can  be  detected  statistically may be
 calculated.
Where: d
       Si
       n
                        = d  Swx/  (1/nj) +  (l/n)~

            the critical value for the Dunnett's procedure
            the square  root of the within mean square
            the common  number of replicates at each concentration
            (this assumes equal replication at each concentration
            the number  of replicates in the control.
14.2.7.8  In this example:
                   MSD = 2.85 (0.158)VTT/2) + (1/2)
                       = 2.85 (0.158M1.0)
                       = 0.450

14.2.7.9  The MSD (0.450) is in  transformed units.   To determine the MSD
in terms of percent survival, carry out the following conversion.

    1.  Add the MSD to the transformed control mean.

                           0.322  +  0.450 * 0.772
                                    113

-------
     2. Obtain the untransformed values for the control mean and the  sum
        calculated in  1.

                          [Sine ( 0.322)  ]2 = 0.1
                          [Sine ( 0.772)  ]2 = 0.487

     3. The untransformed MSD (MSDU) is determined by subtracting the
        untransformed values from 2.

                          MSDU * 0.487  - 0.1 * 0.387

 14.2.7.10  Therefore,  for this  set of data, the minimum difference in
 mean proportion of total mortality between the control  and any SDS
 concentration that can be detected as statistically significant is 0.387.

 14.2.8  Probit Analysis

 14.2.8.1   The data used for the  probit analysis  is  summarized  in
 Table  7.   For the probit analysis,  the SDS concentration  with  100% total
 mortality in  both replicates is considered.  To perform the  probit
 analysis,  run the EPA  Probit Analysis  Program.  Examples  of  the program
 output is  supplied are illustrated  in  Table 8 and Figure  4.

 14.2.8.2   For this example,  the chi-square test for  heterogeneity  was not
 significant.   Thus probit analysis  appears to be appropriate for this set
 of data.
                     TABLE 7.  DATA FOR PROBIT ANALYSIS
                  Control   0.5
                                     SDS Concentration (mg/L)
1.0
2.0
4.0
8.0
Number Dead          2       2       1
Number Exposed      20      20      20
       20
       16
       20
       20
       20
                                    114

-------
 15.   PRECISION  AND  ACCURACY

 15.1  PRECISION

 15.1.1   Single  Laboratory Precision

 15.1.1.1  Single operater, single  laboratory precision data are available
 for eight tests with copper sulfate and five tests with sodium dodecyl
 sulfate.  The data  for the first five tests in Table 6 show that the same
 NOEC  and LOEC, 240  ug Cu/L and 270 ug Cu/U respectively, were obtained
 in all five tests,  which .is the maximum level of precision that can be
 attained.  Three additional tests  (6-8) were performed with narrower
 (20 ug) concentration intervals* to more precisely identify the threshold
 concentration.  The NOEC and LOEC for these tests appear to be 200 ug and
 220 ug Cu/L, respectively.  For sodium dodecyl sulfate, the NOEC'S and
 LOEC'S for all tests were 2.0 and 4.0 mg/l» respectively.

 15.1.2  Multi-laboratory Precision

 15.1.2.1  Data on the multi-laboratory precision of this  test are not yet
available.

15.2 ACCURACY

15.2.1  The  accuracy of  toxicity tests cannot  be determined.
                                   115

-------
TABLE 8.   OUTPUT FROM EPA PROBIT ANALYSIS PROGRAM,  VERSION 3.1,
          USED FOR CALCULATING EC VALUES
    Protdt Analysis of Sheepsnead Minnow Qnbryo-Larval Data
        Cone.
             Number
            Exposed
Control
0.5000
1 . 0000
2.0000
4 . 0000
8.0000
20
20
20
20
20
20
     Number
     Resp.
 Observed
Proportion
Responding
2
2
1
4
16
20
0. 1000
0.1000
0.0500
0 . 2000
0.8000
1.0000
 Adjusted
Proportion
Responding

  0.0000
  0.0174
  -.0372
  0.1265
  0.7816
  1.0000
    Chi - Square Heterogeneity =    0.441
I*i
Sigma
Parameter
Intercept
Slope
0.479736
0.150766
Estimate
1.818003
6.632814


Std. Err.
0.976915 i
1.804695 I


95?, Conf
( -0.096749,
i 3.095611,


idence
3.
10.


L UTll tS
732756)
170017)
    Spontaneous
    Response  Rate
                0.084104
         0.036007
                                             0.013529,
                                        0.154678
          Estimated EC Values and Confidence Limits
Point

EC 1.00
EC 5.00
EC10.00
EC15.00
EC50.00
EC85.00
EC90.00
EC95.00
EC99.00
Cone.

 1.3459
 1.7051
 1.9343
 2.1061
 3,0181
 4.3250
 4.7093
 5.3423
 6.7680
                                          Lower       Upper
                                        95", Confidence Limits
                                         0.4533
                                         0.7439
                                         0.9654
                                         1.1484
                                         2.2676
                                         3.5656
                                         3.8443
                                         4.2566
                                         5.0712
                    1.9222
                    2.2689
                    2.4871
                    2.6523
                    3.6717
                    6.3827
                    7.5099
                    9.6486
                   15.6871
                                      116

-------
          OF ADJUSTED
                           AND
                                       RSGRESSICW
probit
   10+
                     .O
       Figure 4.  Plot of a
djusted  probits and predicted regression line,
                                         117

-------
 TABLE 9.  SINGLE LABORATORY PRECISION OF THE SHEEPSHEAD MINNOW EMBRYO-LARVAL
          SURVIVAL AND TERATOGENICITY TEST PERFORMED IN HW MARINEMIX*
          ARTIFICIAL SEAWATER,  USING EMBRYOS FROM FISH MAINTAINED AND SPAWNED
          IN HW MARINEMIXR ARTIFICIAL SEAWATER AND COPPER  (CU)  AND SODIUM
          DQDECYL SULFATE (SDS)  AS REFERENCE TOXICANTS^,2  3 4  5 6 7



EC!
Test
1
2
3
4
b
fa
7
8
173
t/J
(7)
182
171
17)
(7)
195

EC5
CU
EC10
EC50
(ug/L)
189
17)
(7)
197
187
(7)
(7)
203
198
(7)
" (7)
206
197
(7)
(7)
208
234
(7)
(7)
240
234
(7)
(7)
226
NOEC
LOEC
EC1
(ug/L}(ug/L)
240
240
240
240
240
< 200
220
220
270
270
270
270
270
220
240
240
1.7
(7)
0.4
1.9
1.3



ECS
SUS
EC 10 ECS
(mg/L)
2.0
(7)
0.7
2.2
1.7



2.2 3.1
(7) (7)
0.9 2.5
2.4 3.3
1.9 3.0





0 NOEC LOEC
(mg/L) (mg/L)
2.0
4.0
2.0
2.0
2.0



4.0
8.0
4.0
4.0
4.0



       performed by Terry Hollister, Aquatic Biologist, Houston Facility,
 Environmental Services Division, Region 6, U. S. Environmental
 Protection Agency, Houston, Texas.

2Cyprinodon variegatus embryos used in the tests were less than 20 h
 old when the tests began.  Two replicate test chambers were
 used for the control and each toxicant concentration.  Ten embryos were
 randomly added to each test chamber containing 250 mL of test or control
 water.  Solutions were renewed daily.  The temperature and salinity of the
 test solutions were 24 + PC and 2QO/oo, respectively.

3Copper test concentrations were prepared using copper sulfate.  Copper
 concentrations for Tests 1-5 were: 180, 210, 240, 270, and 300 ug/L.
 Copper concentrations for Test 6 were: 220, 240, 260, 280, and
 300 ug/L.  Copper concentrations for Tests 7-8 were: 200, 220, 240,
 and 280 ug/L.  Tests were conducted over a two-week period.
     concentrations for all tests were: 0.5, 1.0, 2.0, 4.0, and 8.0 mg/L.
 Tests were conducted over a three week period.

5Adults collected in the field.

6For a discussion of the precision of data from chronic toxicity
 tests see Section 4, Quality Assurance.
?Data do not fit the Probit model.
                                        118

-------
 Figure 5.  Data form for Sheepshead minnow embryo-larval  survival/teratogenicity
 test.  Daily record of embryo-larval survival/terata and  test  conditions.
 Test Dates:
 Type Effluent:
 Effluent Tested:

 Original  pH:
     Species:
               Field:
Lab:
Test
Sal im'ty:
     0.0.:
CONCENTRATION:
REPLICATE I
DAYS
#Li ve/Dead
Embryo-Larvae
Terata
Temp. '. °C j
Salinity (ppt)
D.O. (mg/L)'
PH
CONCENTRATION:
REPLICATE I
DAYS
-Live/Dead
Embryo-larvae
Terata
Temp. : <;C
Salinity (ppt)
D.C. (mg/L)
DH

0






I:
0







1







1







2







2







3







3







4







4







5







5







6







6







7







r 7







8







8







9







9






Comments:
Note:F^nal  encpcint for this test is total  mortality (combined  total- number of
dele embryos, dead larvae, and deformed larvae).   See 11.8 and 14.

                                          119

-------
                                 Figure 5.  Continued.
CONCENTRATION:
REPLICATE I
DAYS
#Live/Dead
Embryo-Larvae
Terata
Temp. (°C)
Salinity (ppt)
D.O. (mg/L)
PH
CONCENTRATION;
REPLICATE I
DAYS
#Live/Dead
Embryo-larvae
Terata
Temp. (°C)
Salinity (ppt)
D.O. (mg/L)
PH
II:
0






V:






,

1







1







2







2







3







r 3







4







4







5







"" 5







6







6







7







	 7







8







*







9







9






Comments:
Note:  Fl'nal endpoint for this test is total mortality (combined total  number of
dead embryos, dead larvae, and deformed larvae).   See 11.8 and 14.
                                    120

-------
                                    SECTION 13                       : >-^

                                 TEST METHOD  1,2

         INLAND SILVERSIDE (MENIDIA BERYLLINA) LARVAL SURVIVAL AND GROWTH
                                   METHOD  1006
 1.   SCOPE  AND  APPLICATION

 1.1   This  method  estimates  the  chronic  toxicity of  effluents  and  receiving
 waters  to  the  inland  silverside (Menldla  beryllina),  using  seven-to-eleven-
 day-old larvae in a seven-day,  static-renewal  test*   The  effects  include  the
 synergistic, antagonistic,  and  additive effects of  all  the  chemical,  physical,
 and  biological  components which adversely affect the  physiological  and
 biochemical functions of the test  species.

 1.2   Detection  limits of the toxicity of  an effluent  or pure  substance are
 organism-dependent.

 1.3   Single or  multiple excursions  in toxicity  may  not  be detected  using  24-h
 composite  samples.  Also, because of the  long sample  collection period
 involved in composite sampling,  and because the  test  chambers are not sealed,
 highly  volatile and highly  degradable toxicants  in  the  source may not be
 detected in the test.

 1.4   This method  should be  restricted to  use by,  or under the supervision of,
 professionals experienced in aquatic toxicity testing.

 2.  SUMMARY OF METHOD

 2.1   Seven-to-eleven-day-old larvae are exposed  in a  static-renewal system for
 seven days to different concentrations of effluent or to receiving water.
Test  results are based on the survival and growth {increase in weight) of the
 larvae as compared to the control.

3.  DEFINITIONS                       "'' '**

   (Reserved for addition of terms at a later date).

4.  INTERFERENCES                                         ^tf

4.1  Toxic substances may be introduced by contaminants in dilution water,
glassware,  sample hardware,  and testing equipment {see Section 5,  Facilities
and Equipment).

TThe format used for this method was taken from Kopp,  1983.
2This method was adapted from Heber, Hughes,  Schimmel, and Bengtson, 1987,
Environmental  Research Laboratory, U.  S. Environmental Protection  Agency,
Narragansett,  Rhode Island.

                                      121

-------
 4.2  Adverse effects of low dissolved oxygen concentrations (DO),  high
 concentrations  of suspended and/or dissolved solids,  and  extremes  of pH,  may
 mask or confound the effects of toxic substances.

 4.3  Improper effluent sampling and handling may adversely affect  test  results
 (see Section 8,  Effluent and Receiving Water Sampling and Sample Handling).

 4.4  Pathogenic  and/or predatory organisms  in the  dilution water and effluent
 may affect  test  organism survival,  and confound  test  results.

 4.5  Food added  during the  test may sequester metals  and  other  toxic
 substances  and  confound test results.

 5.   SAFETY

 5.1   See Section 3,  Health  and  Safety.

 6.   APPARATUS AND  EQUIPMENT

 6.1  Facilities for holding  and  acclimating  test  organisms.

 6.2   Brine  shrimp culture unit  —  see  7.16  below.

 6.3   Menidia  beryllina  culture  unit  --  7.17  below, Middaugh, 1985, and
 Middaugh, et  al.,  1986,  1987 for detailed culture methods.  This test requires
 from 180 to  360  seven-to-eleven-day-old larvae.  It is preferable to obtain
 the  test organisms from  an  inhouse culture  unit.  If  it is not feasible to
 culture fish  inhouse,  embryos or  larvae can  be obtained from other sources by
 shipping them in well  oxygenated saline water  in insulated containers.

 6.4   Samplers — automatic  sampler,  preferrably with sample cooling
 capability,  that can collect a  24-h  composite sample of 5 L.

 6.5   Environmental chamber  or equivalent facility with temperature control
 (25+  2°C).

 6.6   Water purification  system  — Millipore Super-Q, Deionized water (DI)  or
 equivalent.

 6.7   Balance, analytical — capable of accurately weighing to 0*0001 g.

6.8  Reference weights, Class S -- for checking performance of balance.
Weights should bracket the expected weights of the weighing boats and the
expected weights of the weighing boats plus fish.

6.9   Drying  oven -- 105°C, for drying larvae.

6.10  Air pump -- for oil free air supply.

6.11  Air lines, plastic or pasteur pipettes, or air stones, --  for gently
aerating water containing the fragile larvae and or supplying air to test
solution with low DO.

                                       122

-------
6.12  pH and DO (non-stirring probe) meters — for routine physical and
chemical measurements. Unless the test is being conducted to specifically
measure the effect of one of the above parameters, a portable, field-grade
instrument is acceptable.

6.13  Standard or micro-Winkler apparatus -- for calibrating DO (optional).

6.14  Desiccator — for holding dried larvae.

6.15  Light box — for counting and observing larvae.

6.16  Refractometer -- for determining salinity.

6.17  Thermometers, glass or electronic, laboratory grade — for measuring
water temperatures.

6.18  Thermometers, bulb-thermograph or electronic-chart type — for
continuously recording temperature.

6.19  Thermometer, National Bureau of Standards Certified (see USEPA METHOD
170.1, USEPA, 1979) — to calibrate laboratory thermometers.

6.20  Test chambers —  four (minimum of three) chambers per concentration.
The chambers should be borosilicate glass or non-toxic disposable plastic
labware.  To avoid potential contamination from the air, the chambers should
be covered during the test.

6.20.1  Each test chamber for the  inland silverside should contain a minimum
of 750 ml of test solution. A modified Norberg-Mount (1985) chamber
(Figure 1), constructed of glass and silicone cement, has been used
successfully for this test.  This  type of chamber holds an adequate column
of test solution and  incorporates  a sump area from which test solutions can
be siphoned and renewed without disturbing the fragile inland silverside
larvae.  Modifications for the chamber are as follows:  1) 200 urn mesh nylon
screen  instead of stainless steel  screen; and 2) thin pieces of glass rods
cemented with silicone to the NYLONR screen to reinforce the bottom and
sides to produce a sump area in one end of the chamber.  Avoid excessive use
of silicone, while still ensuring  that the chambers do not leak and the
larvae  cannot get trapped or escape into the sump area.  Once constructed,
check the chambers for leaks and repair if necessary.  Soak the chambers
overnight in sea water (preferably in flowing water) to cure the silicone
cement  before use.  Other types of glass test chambers, such as the 1000 ml
beakers used in the short-term sheepshead minnow  larval survival and growth
test, may be used.  It is recommended that each chamber contain a minimum  of
50 ml per larvae and  allow adequate depth of test solution  ( 5.0 cm).

6.21  Beakers — six  Class A, borosilicate glass  or non-toxic plasticware,
 1000 ml for making test  solutions.

-------
          9cm
GLASS
REINFORCEMENTS
9cm
                                                SUMP

                                          from
                     F    H
                     From: Heber, Hughes, Schimmel, and Bengtson,

-------
6.22  Mini-Winkler bottles — for dissolved oxygen calibrations.
6.23  Wash bottles — for deionized water, for washing embryos from substrates
and containers, and for rinsing small glassware and  instrument electrodes and
probes.
6.24  Crystallization dishes, beakers, culture dishes, or equivalent — for ,
incubating embryos.
6.25  Volumetric flasks and graduated cylinders — Class A, borosilicate glass
or non-toxic plastic labware, 10-1000 ml for making  test solutions.
6.26  Separatory funnels, 2-L — Two-four for culturing Artenna.
6.27  Pipets, volumetric — Class A, 1-100 ml.
6.28  Pipets, automatic — adjustable,   1-100 ml.
6.29  Pipets, serological —  1-10 ml_, graduated.
6.30  Pipet bulbs  and fillers — PROPIP£TR, or equivalent.
6.31  Droppers, and glass tubing with fire polished  edges,  4mm  ID  — for
transferring  larvae.
6.32  Siphon with  bulb  and clamp — for  cleaning  test  chambers.
6.33  Forceps — for transferring dead  larvae  to  weighing  boats.
6.34  NITEXR mesh  sieves  (<  150  urn and  500 urn) — for  collecting Artjanla  and
fish  larvae.
7.  REAGENTS AND CONSUMABLE MATERIALS       :. ^,  •
7.1   Sample  containers  — for sample  shipment  and storage  (see  Section 8,
Effluent and  Receiving  Water  Sampling  and  Sample  Handling).
7.2   Data sheets  {one  set per test)  —  for data recording  (Figures 7,  8,  and
9).
7.3   Tape, colored —  for  labelling  test chambers
7.4   Markers,  water-proof  --  for marking containers, etc,
 7.5   Vials,  marked —  24/test,  containing 4% formalin or 70% ethanol,  to
 preserve larvae.  (Optional).
 7.6   Weighing boats,  aluminum  — 26/test (2 extra).
                                        125

-------
 7.7  pH buffers 4, 7, and 10 (or as per instructions of instrument
 manufacturer) for standards and calibration check (see USEPA Method 150.1,
 USEPA,  1979).

 7.8  Membranes and solutions for dissolved oxygen probe (see USEPA Method
 360.1,  USEPA, 1979),  or reagents for modified Winkler analysis.

 7.9  Laboratory quality assurance samples and standards for the  above methods.

 7.10  Reference toxicant solutions (see Section 4,  Quality Assurance).

 7.11  Formalin (4%)  or 70% ethanol for use as a preservative for the fish
 larvae.

 7.12  Reagent water  — defined  as distilled or deionized water that does  not
 contain  substances which are toxic to the test organisms (see paragraph 6.6
 above).

 7.13  Effluent,  surface water,  and dilution water —  (see Section  7,  Dilution
 Water, and  Section 8,  Effluent  and surface Water Sampling and Sample  Handling),

 7.13.1   Saline test and dilution  water --  The salinity of the test  water  must
 be  in the range  of 5  to 32 o/00.   The salinity should  vary by no more than
 + 2  o/oo among the chambers  on  a  given day.   If effluent  and  receiving water
 tests are conducted concurrently,  the salinities of these tests  should be
 similar.

 7.13.2  The overwhelming majority of  industrial  and sewage  treatment effluents
 entering marine  and estuarine systems contain little or  no  measurable salts.
 Exposure of Menidia beryllina larvae  to  these effluents will  require
 adjustments in the salinity  of  the  test  solutions.  It  is  important to
 maintain a  constant salinity across all  treatments.  In addition,  it may  be
 desirable to match the  test  salinity  with  that  of the receiving water.
 Hypersaline brine  (100  °/oo) derived  from  natural seawater  may be used to
 adjust the  salinities.   However,  the  use of  hypersaline brine will  limit  the
 concentration  of effluent  that  be tested to  70%  at 30 o/00  salinity and 80%
 at 20 o/oo  salinity.

 7.13.3  Hypersaline brine:   Hypersaline brine  (HSB) has several  advantages
 that make it desireable for  use in toxicity  testing.  It can be made from any
 high quality, filtered  seawater by evaporation,  and can be added to the
 effluent or to deionized water to increase the  salinity.  HSB derived from
 natural seawater contains the necessary trace metals, biogenic colloids,  and
 some of the microbial components necessary for adequate growth,  survival,
 and/or reproduction of marine and estuarine organisms, and may be stored  for
 prolonged periods without any apparent degradation.

 7.13.3.1  The  ideal container for making HSB from natural seawater is one  that
 (1)  has a high surface to volume ratio, (2)  is made of a non-corrosive
material, and (3) is easily cleaned (fiberglass  is ideal).  Special care
should be used to prevent any toxic materials from coming in contact with
                                       126

-------
                             .t      e
      Teach any substances that would cntaminate
   method used is a thermostatically
                                                        -eria15 d° not corrode
                                                              °ne Sljc«ssful
                                                           th'™*"* "«"
        be in direct  contact  with  the brine   ^ ™1 "y ?^her mat^ials that
   detergent  should be  used,  followed by sweral9??? ?UalJty bfode9^dable
   deionized  water rinses   Hioh n»ai,-?f fev5ral (at 'east three  thorough
   seawater should be ff tere^to^t LL '?n pre:e:ab^ ^igh  salinity)
   generator.  Water should be collect!  on an"?,,^ P'aCfng  into «e  brine
   possibility of contamination             "" lncommi'19 tide  to minimize  the
  Tncrease water evaporation.  The brfne
 volume being generated) to ensure tha?
 and that the temperature does not
 added to the brine to obtain the
                                                                   and to
                                                                Depending on
                                                      -Ot   Ceed 10° °/0°
                                                                         be
 portable  containers.  Twenty-Liter
 jugs  are  suitable.  The containers «
 date  the  brine was generate  an^its
 stored ,„ the dark and maintain^
                                                         dl>6Ctly  into
                                                   Polycarbonate water cooler

                                                         'abe"ed W1'th the
                                                                  «h»uTd be
and brine before mixing in the effluent
                                                         the de1°"<»d water
                                                         test
                                                                      the
one part in f ,„. (one pat brineto   our
HSB needed to make 1  L of seawter
of deionized water required
                                       ^u"1^  fr« a HSB of  '«»
                                         w*'  2°° "t'  i$ the 1uant"* of
                                        deference, 800 ml, is the  uant
                                                                   quantity
                                    127

-------
Tabl! V]lus^ates the composition  of  test  solutions at
                                                                     20
                                                           -

  7.14  ROTIFER CULTURE

?KdeSA °r harPac<1co1d c°P^POds that 'mayhav^ ^ been  Ldvertently
introduced can rap.dly take over the cuHure.   If this  occurs^  d?s^rd  the



7.15  ALGAL CULTURES


7.15.1  Tetraselmus  suecica or  Chlorella sp. (see Middauah  et al   ^K7\
                                     128

-------
TABLE 1.

  Effluent
 Concentration
    (X)
Volume of
Effluent
(0 o/oo)
   (ml)
Volume of
Deionized
 Water
  (ml)
 Volume of
Hypersaline
  Brine
32
10
3.2
1.0
0.32
Control
960
300
98
30
9.6
--„
1440
2100
2302
2370
2390
2400
                                                                 600

                                                                 600

                                                                 600

                                                                 600

                                                                600

                                                                600
                                 129

-------
  7-15.2  Formula for algal  culture
                                nutrients
                                    Mir
                                                The solution
Approxlmately 6  - 8 continuous cu?tu«sw???ry SeC°nd day basfs*
of/our 12-L rotifer cuitn"""  cultures wl17 """** *-u-
7.16  BRINE SHRIMP

7.16.1
                          NAUPU; (see P.,tfep anrf


'.:

                                       ' «•
                                          "°ltort»

                                                     arc
7.16.2

    1.



   2.




   3.
          a nauplii are obtained as follows:
     •t 270C.     tn
     the geographic strain of
                                               a
-------
         4.   Drain  the
             and rinse
                                 ?„"
                                                           «»n ,o mln without
                                                          a.-
                                     3r equivalent

    7-16.3  Testing Artenjla nauplii as  food for bioassay organisms.
                                                      of  each  new  supply Of brfne
       the inland silverside larvae    The  1«™J     Trt 9°od S«rv1val and growth
    of the brine shrimp  naupl  frost  bJ of  the  < Jf*  t0 evaluate  the sultablliS
    and stage  of development9 a   those  used  r2ut?^v9-°9^phtcal °rt9fn' sP
    rep hcate  test  vessels,  each conta ining 1   arJae        °XiCf   tests'
    will  Prov.de  sufficient  data to detect^l^cls
   snould be the sa.e as
                                                  v?sse's  •"<  volume  of  control
                                                              Start of the fit.
        is  acceptable  if there are no
   the  surv.al and growth of the
                                                         f
                                                              'te5t-  The
   7-16.4  Use of Artemia nauplii as food for Menidia larvae
         survival and growth  test.   Equal  amoiin
  replicate test chamber  to minim ze the var?ab   itv
  Suffident numbers  of nauplii  shoulri h«  flJ      *
  overnight in  the  test chambe s   A  adequate
  be  provided to  e
                                                           throu9hout the 7-

                                                            T&t b* fed to
                                                        arval weight.

                                     adequate               ^ remain a'1ve
  be provided to each replicate on a dai?v basis * ?°L?XCeSSTVe amount s^"ld
  naupln will result in a depletion in DO tn hli     dl"9 excess've amounts  of
  3.5 mg/L)   As much of the Snlaten         as oossihTf ^Pt?Sle leve'  (below
                                                  slble<.ushould  »e siphoned from
                                                       that the  larvae
 each chamber prior to test    uon
 principally eat newly hatched nauplii
 7.17  MBODA
                         BROOD STOCK

 atherinid species «edor
                                                  is one of three species in
                                                              a"d
west to Vera Cruz,  Mexico  Johnson   1975)     t
from the freshwaters
                                                               T7oFida and

-------
   (Chernoff et al    1981)  to hypersaline lagoons (Simmons  1957)


   the circulation of water in the culture tSnk,  ?M&.   t™!  lnte"-uption in
                                                                         Th'

7.17.2  Menidia
culture methods)
be
                           adults {see Middauah  et al    10*7   t    „«....
                           lture in thl TK   I       '  ]9S7'  for  detailed

                                                                   '
                                                                         ,„
 generally spawn fw 4-6  mSs                   fter 7"2 m°nths and wil1

                                                                       the
easTTy-TTTthe field  on  that  basis            M)> and can be
                                    132

-------

s rs,  r      ,'«
  24-h old.
                                   T can be increased from 12-h  old to

 Artemia nauplii  (8 - ?S dais  of
 ^UTTTe 9 ^  ,1  ays  ?d w^en the
 and reared in the ?est ,.60^
                                                        -°»«
                                                be feedir|9 wel)
                                                                of

                       ^
 8.   EFFLUENT AND RECEIVING
 8.1  See Section 8.
 9-   CALIBRATION  AND STANDARDIZATION
 9.1  See Section  4, Quality Assurance.
 10.  QUALITY  CONTROL
 10.1   See  Section 4, Quality Assurance.
 11.   TEST  PROCEDURES
11.1  TEST SOLUTIONS
11.1.1  Surface Waters
                      toxicity  is
                                   133
                                         with samples used directly as

-------
  11.1.2  Effluents

 •11.1.2.1  The_selection of the effluent test concentrations should be based
 on the objectives of the study.  One of two dilution factors, approximately
 0.3 or 05, is commonly used.  A dilution factor of approximately 0.3 allows

 (10Wn93WtWlW ]3? ± meffiUent USi"i ?nly fl*Ve e?fluent Concentration
 1100*, 30%, 10%, 3%, and 1%).  However, if hypersaline brine is used to
 adjust salinities, the maximum effluent concentration will be 80% at
 20 o/oo salinity, and 70% at 30 o/00 salinity.  This series of dilutions
 minimizes the level of effort, but because of the wide interval between test
 concentrations provides poor test precision.  A dilution factor of 0 5
 provides greater precision, but requires several additional dilutions to
 span the same range of effluent concentrations.   Improvements in precision
 decline rapidly as the dilution factor is increased beyond 0.5.

 11.1.2.2  ]f the effluent is known or suspected  to be highly toxic,  a lower
 range of effluent  concentrations  should be used, beginning at  10%.   If a
 h]9h  rate of mortality is observed during the  first 1  to 2 h of  the  test,
 additional  dilutions at the lower range (3%,  1%, 0.3%,  and 0.1%)  of  effluent
 concentrations should be added.

 1T.1.2.3  The  volume of effluent  required  to start the  test and  for  daily
 renewal  of  four replicates  per  treatment,  each containing 750 ml  of  test
 solution,  is approximately  5  L.   Prepare  enough  test  solution at  each
 effluent concentration  to provide 400 ml  additional volume for chemical
 analyses.

 11.1.2.4  Tests  should  begin  as soon as possible after  sample collection,
 preferably within 24  h.   If the persistence of the  sample toxicity is not
 known, the maximum holding  time should not exceed  36 h  for  off-site toxicity
 studies.  In no case  should the test be started more than 72 h after  sample
 collection.  Just prior to  testing, the temperature of  the  sample should be
 adjusted to  (25 + 2<>C)  and maintained at that temperature until portions
 are added to the dilution water.

 11.2  START OF THE TEST

 11.2.1  M. beryl!ina  larvae seven to 11 days old can be used to start the
 survival and growth test.  At this age, the inland silverside feed on
 newly-hatched Artemia nauplii.  At 25<>C, tests with M. beryllina can be
 performed at salinities ranging from 5 o/oo to 32 o/00.   If the test
 salinity ranges from 16 to 32 o/00t the salinity for spawning, incubation,
 and culture of the embryos and larvae should be maintained within this
 salinity range.  If the test salinity is in the range of 5 o/00  to
 15 o/oo, the embryos may be spawned at 30 o/00, but egg incubation and
 larval rearing should be at the test salinity.   If the specific  salinity
required for the test differs from the rearing  salinity, adjustments  of
5 o/oo daily should be made over the three days prior to start of test.
                                     134

-------
     1.2.2  One Day Prior to  Beginning of Test.
             ISMS £
                                              bath'
                                                            nauplii wil1  be
                                                            ^cutator  to the

 dish, 3 - 4 cm in  diameter, or Mal  nin^tf    ThP trl
 readily escape from a pipe te   ?r  sfer the     '
                                                                   1ization
                                      .
11.3  LIGHT,  PHOTOPERIOD, SALINITY,  AND  TEMPERATURE

11.3.1

salinity should be in the range of 5
                                                                The test
                                    135

-------
   salinities of these tests should be similar.  The water temperature in th*
   test chambers should be maintained at  25 + 2K.         temperature in the


   11.4  DISSOLVED OXYGEN CONCENTRATION (DO)
onlv
only
            i         + a!fect.the  toxicity of effluents and should be used
          a last resort to maintain satisfactory DO.   The DO should be

             ss asa: s           --             ~
  11.5  FEEDING



  11.5.1   Artemia nauplii are prepared as described above.


  11.5.2   The test larvae are fed newly-hatched {less  than 24-h-old)

  naupln  once a day from Day 0  through Day 6; larvae  are not fed on

  Equal Amounts of Artemia must  be fed" to eac  repHcate test
   P=.«
                                                        .
 ensure that the  larvae principally eat  newly hatched nanplii


 11.5.3  On  days  0, 1, and 2, transfer 4 g wet weight or 4 mL  of
 concentrated,  rinsed Artemia nauplii to seawater in a 100 ml  beaker  and

             hlUme °t--° ^   ^ate °r  SW1>1 the suspension to euaily
               -aupjn "hl1e W1thdraw1"9 individual 2 ml  portions of the
         suspension by pipette or adjustable syringe to transfer to each
 replicate test chamber.   Because the nauplii will  settled concentrate at

 I  ?h^P  °f th^ ? P6tte durl'"9 the fansfSr.  limit  the volume Sf concentrate
 withdrawn each time to a  2-mL portion for one  test chamber helps ensure an
 equal distribution to the replicate chambers   Equal  distribution of food to
 the replicates is critical for successful  tests.          '"""on or rood to


        °I! D?ys 3"6'  transfe'- 6 9 wet weight or 6  ml of the  Artemia
           th! lar!ae Surv1val ^te in  any replicate on any day falls below

        K   thehv°luf °f Artem
-------
      "•6   DAILY  CLEANING OF TEST CHAMBERS
                       or
                                                         - -a
                                                             fl'tted

    . 7
         TEST SOLUTION RENEWAL
                           a

            have proven suitablefor  effi,«?
on-site toxicity studies no  more  thin 24
of the effluent and use in a  t°M{y t'i?
                                                     t.,t is stored ,„ an
                                              nta'ers Such as 8-20
                                                          and storage.  For
                                                                ^
            one solution is aerted  the   u   OXysfn co"ce"t^tion
 aerated.  Aerate the test solution 'in ?h- tl,50nh8nJrations must be
 larvae are not disturbed.              the test  Cambers gently, so that

 H.8  ROUTINE CHEMICAL AND  PHYSICAL ANALYSIS


(Figure 7*):'  •?ni""'  the f«"^"9 -asuren,ents are made and  recorded
                                     137

-------
11.8.1.1  DO is measured at the beginning and end of each 24-h exposure
period in one test chamber at all test concentrations and in the control.

11.8.1.2  Temperature, pH, and salinity are measured at the end of each 24-h
exposure period in one test chamber at all test concentrations and in the
control-  The pH is measured in the effluent sample each day.

11.9  OBSERVATIONS DURING THE TEST

11.9.1  The number of live larvae in each test chamber are recorded daily
(Figure 7), and the dead larvae are discarded.

11.9.2  Daily test observations, solution renewals, and removal of dead
larvae, should be carried out carefully to protect the larvae from
unnecessary disturbance during the test.  Care should be taken to see that
the larvae remain immersed at all times during the performance of the above
operations.

11.10  TERMINATION OF THE TEST

11.10.1  The test is terminated after seven days of exposure.  At
termination, the number of surviving larvae in each test chamber are counted
and as a group are immediately prepared for drying and weighing, or are
preserved in 4% formalin or 70% ethanol for drying and weighing at a later
date.  For immediate drying and weighing, siphon or pour live larvae onto a
500 urn mesh screen in a large beaker to retain the larvae and allow Artemia
to be rinsed away.  Rinse the larvae with cold deionized water to remove
salts that might contribute to the dry weight.  Sacrifice the larvae in an
ice bath of deionized water.   Small aluminum weighing boats can be used to
dry and weigh larvae.  An appropriate number of aluminum weigh boats (one
per replicate) are marked for identification and weighed to 0.01 mg, and the
weights are recorded (Figure 8) on the data sheets.

11.10.2  Immediately before drying, the preserved larvae are rinsed in
distilled water.  The rinsed larvae from each test chamber are transferred,
using forceps, to a tared weighing boat and dried at 6QQC for 24 h, or at
105°C for a minimum of 6 h.  Immediately upon removal from the drying
oven, the weighing boats are placed in a desiccator to cool and to prevent
the adsorption of moisture from the air until weighed.  Weigh all weighing
boats containing the dried larvae to 0.01 mg, subtract the tare weight to
determine dry weight of larvae in each replicate, and record (Figure 8) of
data sheets).  Divide the dry weight by the number of larvae per replicate
to determine the average dry weight, and record (Figures 8 and 9) of the
data sheets.  Complete the summary data sheet (Figure 9) after calculating
the average measurements and statistically analyzing the dry weights and per
cent survival for the entire test.
                                      138

-------
  12.  ACCEPTABILITY OF TEST RESULTS

  12.1  Test results are acceptable if  (1) the average survival of control
  larvae is equal to or greater than 80#, and (2) where the test starts with
 7-day old larvae, the average dry weight of the control larvae, when dried
  immediately after test termination, is equal to or greater than 0.50 mg, or
 the average dry weight of the control larvae preserved in 4% formalin or 70%
 ethanol is equal to or greater than 0.43 mg.

 13.  SUMMARY OF TEST CONDITIONS

 13.1  A summary of test conditions is listed in Table 2.     %

 14.  DATA ANALYSIS

 14.1  General

 14.1.1   Tabulate and  summarize the data.

 14.1.2   The  end points  of toxicity tests  using  the inland  silverside are
 based  on  the adverse  effects  on  survival  and growth.   Point  estimates*  such
 as LCI,  LC5,  LC10  and LC50, are  calculated  using  Probit Analysis  (Finney,
 1971).   LOEC and NOEC values,  for survival  and  growth, are obtained  using  a
 hypothesis test approach  such  as Dunnett's  Procedure  (Dunnett,  1955)  or
 Steel's Many-one Rank Test  (Steel,  1959;  Miller,  1981).  See  the  Appendix
 for examples of the manual computations,  program  listings, and  examples  of
 data input and  program  output.

 14.1.3  The  statistical tests  described here must  be used  with  a  knowledge
 of the assumptions upon which  the tests are contingent.  The  assistance  of a
 statistician  is recommended for  analysts  who are not proficient in
 statistics.

 14.2 EXAMPLE OF ANALYSIS  OF MENIDIA SURVIVAL DATA

 14.2.1  Formal  statistical analysis of the  survival data is outlined  in
 Fiqure 2.  The  response used in  the analysis is the proportion of animals
 surviving in each test or control chamber.   Separate analyses are performed
 for  the estimation of the NOEC and LOEC end points and for the estimation of
 the  LCI, LC5, LC10 and LC50 end points. Concentrations at which there is no
 survival in any of the test chambers are excluded from statistical analysis
 of the NOEC and  LOEC, but included in the estimation of the LC end points.

 14.2.2  For the  case of equal  numbers of replicates across all
 concentrations  and the control, the evaluation of the NOEC and LOEC end
 points is made  via a parametric test, Dunnett's Procedure,  or a
 nonparametric test, Steel's Many-one Rank Test,  on the arcsin transformed
 data.  Underlying assumptions  of Dunnett's Procedure,  normality and
 homogeneity of variance, are formally tested.  The test for normality is the
 Shapiro-Wilks Test, and  Bartlett's Test is used to determine the homogeneity
 of variance.   If either  of these tests fail, the nonparametric test,  Steel's
Many-one Rank Test, is used to determine the NOEC and  LOEC  end points.  If
the assumptions of Dunnett's Procedure are met,  the end points are estimated
by the parametric procedure.
                                      139

-------
  TABLE 2.   SUMMARY OF RECOMMENDED TEST CONDITIONS FOR  THE  INLAND  STI VFpqrnr
            (MENIDIA BERYLLINA)  LARVAL  SURVIVAL  AND GROWTH  TEST    SILVERSIDE
    1.  Test type:
    2.  Salinity:

   3.  Temperature:
   4.  Light quality:
   5.  Light intensity:
   6.   Photoperiod:
   7.   Test  chamber  size:
  8.   Test  solution  volume:
  9.  Renewal of test
       concentrations:
 10.  Age of test organisms:
 11.  Larvae/test chamber
       and control:
 12.   Replicate
       chambers/concentration
 13.   Source of food:
 14.   Feeding regime:
15.  Cleaning:

16.  Aeration:
   Static-renewal
   5  o/oo  to  32  o/oo  (+  2  o/oo  of
   the  selected  test  salinity)
  .25 + 20C
  Ambient laboratory illumination
  10-20 uE/m2/s (50-100 ft-c) (ambient
  laboratory, levels)
  14 h  light,  10 h darkness
  300 mL  - 1  L  containers
  250-750  mL/replicate  (loading  and
  DO  restrictions  must be  met)
  daily
  7-11 days post hatch
  15 (minimum of 10)
 4 (minimum of 3)
 Newly hatched Artemia  nauplii
 Feed  0.10  g  wet weight Artemia
 nauplii  per  replicate orTdays 0-2;
 Feed  0.15  g  wet weight Artemia
 nauplii  per  replicate on~dayT"3-6
 Siphon daily,  immediately before test
 solution renewal and feeding
None* unless DO concentration falls
below 60% of saturation,  then
aerate all  chambers.  Rate should be
less than 100 bubbles/min.

-------
                                        H»i iiril HMIiiii nil t«tl iiiiit'ifirillna-l
TABLE 2.  SUMMARY OF RECOMMENDED  TEST CONDITIONS FOR THE INLAND SILVERSTDF
                   BERYLLINA)  LARVAL  SURVIVAL AND GROWTH TEST (CONTINUED)
 17.  Dilution water:

 18.  Effluent concentrations
 19.  Dilution factor:
 20.  Test duration:
 21.   Effects  measured:
Uncontaminated  source of sea
water or  deionized  water mixed
with hypersaline  brine.
At least  5 and  a  control
Approximately 0.3 or  0.5
7 days
Survival and growth (weight)
                                   141

-------
     14-2.4  Probit Analysis  (Finnev  10711  -

TABLE 3.  INLAND SILVERSIDE SURVIVAL
                                                            DATA
   RAW
     0.80
     0.87
     0.93
0.73
0.80
0.87
0.80
0.33
0.60
0.40
0.53
0.07
0.0
0.0
0.0
0.0
0.0
0.0
ARC SINE
TRANS-
FORMED
                           1.107
                           1.202
                           1.303
               .024
              1.107
              1.202
 0-612
 0.886
                 0.685
                 0.815
                 0.268
 MEAN (7,-
 S-,-2
    1.204
    0.010
    1
1.111
0,008
2
0-868   0.589
0-061   0.082
3       4
 14.2.6   Test  for  Normality
observations by sub trying the ^   oT.l?^"* 1s t0 cente' the
concentration from each observation in till °bservatio^ within a
observat7ons are summarized in Tab?e 4    * concent™tion.  The centered
                                    142

-------
                                SURVIVAL
                                 SURVIVAL DATA
                             PROPORTION SURVIVING
                              „   ARCSIN
                              TRANSFORMATION
LC1PL°C5.TLCE10TILMCA5T0E
                                                           DISTRIBUTION
                NORMAL  DISTRIBUTION
                             BARTLETT'S  TEST
                                                  HETEROGENEOUS
                                                     VARIANCE
   HOMOGENEOUS  VARIANCE
                 EQUAL NUMBER OF
                   REPLICATES?
                                    EQUAL NUMBER OF
                                      REPLICATES?
     T-TEST WITH
     BONFERRONI
                         STEEL'S MANY-ONE
  WILCOXON RANK SUM
      TEST WITH
BONFERRONI ADJUSTMENT
     ADJUSTMENT
"gun, 2.  How chart for statistical  analysis  of Menida survival data.

-------
     XH-
   — UJU.
*-coo<


*~- uioru.
O(/)0.—
                                 iO
                                 o    6
                       NOlldOdOdd
T

d
T
q
d
                                   144

-------
        TABLE 4.  CENTERED OBSERVATIONS FOR SHAPRIO-WILKS EXAMPLE
Effluent Concentration (%} ,^:>:>'
Replicate
A
B
C
Control
-0.097
-0.002
0.099
1.0
-0.087
-0.004
0.091
3.2
0.239
-0.256
0.018
10.0
0.096
0.226
-0.321
14.2.6.2  Calculate the denominator,  D9  of the statistic:

                          n
D -
                                  X)2
    Where:   Xj = the ith centered observation
            X  = the overall mean of the centered observations
            n  = the total number of centered observations
14.2.6.3  For this set of data,    n =.12
                                   t? .    i
                                           (0.002) = 0.0002
                                        12

                                   D = 0.3214

14.2.6.4  Order the centered observations from smallest to largest
where x(i) denotes the ith ordered observation.  The ordered
observations for this example are listed in Table 5.
  TABLE 5.  ORDERED CENTERED OBSERVATIONS FOR SHAPIRO-WILKS EXAMPLE
i
1
2
3
4
5
6
xd>
-0.321
-0.256
-0.097
-0.087
-0.004
-0.002
i
7
8
9
10
11
12
xtD
0.018
0.091
0.096
0.099
0.226
0.239
                                     145

-------
  14.2.6.5  From Table 4, Appendix B, for the number of observations, n,
  obtain the coefficients ai, 32, ... a^ where k is approximately
  n/2. For the data in this example, n = 12 and k = 6.  The a-,- values are
  listed in Table 6.

:  14.2.6.6  Compute the test statistic, W, as follows:
i
                         k
               w s
                   D
The differences
in this example,
               W =
                                        - Xti)) ]2


                           - x(i) are listed in Table 6.  For the data


                                      = °-945
  TABLE 6.  COEFFICIENTS AND DIFFERENCES FOR SHAPIRO-WILKS EXAMPLE
                    a
                              xtn-i+1) -
1
2
3
4
5
6
0.5475
0.3325
0.2347
0.1586
0.0922 .
0.0303
0.560
0.482
0.196
0.183
0.095
0.020
XH2) - Xd)
x\ 1 1 ) - x'^ )
X(10) - X(3)
x(9) - x^4)
x(8) - x(5)
XC7) . X 6
   14.2.6.7   The  decision rule for this test  is to compare W as calculated
   in  14.2.6.6  to a critical value found  in Table 6, Appendix B.   If the
   computed W is  less  than  the critical value, conclude that the data are
   not  normally distributed.  For the  data in this example, the critical
   value  at a significance  level of 0.01  and  n - 12 observations is 0.805.
   Since  W =  0.945  is  greater than the critical value, conclude that the
   data are   normally  distributed*

   14.2.7 Test for Homogeneity  of Variance

   14.2.7.1   The  test  used  to examine  whether the variation in  survival  is
   the  same across  all effluent  concentrations  including  the control, is
   Bartlett's Test  (Snedecor and Cochran, 1980).  The  test  statistic  is  as
   follows:
                    P
                [  (  I
                   1=1
                            In S2 - t YJ  In
                                   1=1
                                       146

-------
    Where
          P  =
degrees of freedom for each effluent concen^
tration and control, Vi = (nj - 1)

number of levels of effluent concentration
including the control
                    t.E,Vi  Si2j
                                 P          P
          C  = 1  + [ 3(p-l)]-l  [  I 1/Vi  - (  t
          In = loge

          i   -1,2, ..., p where p is the number of concentrations
                            including the control
          nj = the number of replicates for concentration i.

14.2.7,2  For the data in this example, (See Table 3) all effluent
concentrations including the control have the same number of  replicates
{ni - 3 for  all i).  Thus, Vi - 2 for all i.

14.2.7.3  Bartlett's statistic is therefore:
       B -  [(8)ln(0.0402) - 2 I ln(S?j/1.2083
                              i = l     1

         =  [8(-3.2139) - 2(-14.73T)J/1.2083

         -  3.7508/1.2083

         =.  3.104

14.2.7.4  B is approximately distributed as chi-square with p - 1  degrees
of freedom, when the variances are in fact the same.  Therefore, the
appropriate critical value for this test, at a significance level  of 0.01
with three degrees of freedom, is 11.345.  Since B = 3.104 is less than
the critical value of 11.345, conclude that the variances are not
different.
                                     147

-------
14.2.8  Dunnett's Procedure

14.2.8.1  To obtain an estimate of the pooled variance for the Dunnett's
Procedure, construct an ANOVA table as described in Table 7.

                           TABLE 7.  ANOVA TABLE
Source
Between
Within
Total
Where:
df
P - 1
N - p
N - 1
p = number
N = total
n-j = number
Sum of Squares Mean Square(MS)
(SS) CSS/df)
2
SSB SB ^ 5SB/(p-l)
2
ssw ' SW - SSW/(N-p)
SST
of effluent concentrations including the contn
number of observations n] + p? ..* +np
of observations in concentration i
SSB - I Tj2/ni - G2/N
                                          Between Sum of Squares
           SST «
                               Total Sum of Squares
           SSW = SST - SSB
                               Within Sum of Squares
                                                                 r
            G   = the grand total of all  sample observations, G = I T-j

            T-j  = the total of  the replicate measurements for
                 concentration  "i"
            (ii  = the jth  observation  for concentration  "i"  (represents
                 the proportion surviving for effluent  concentration
                 i  in  test chamber  j)
                                     148

-------
14.2.8.2  For the data  in this example

    __ ^ 	 __ _ __           -•*
    ri 1 — no — no ~ n/i — ^
    N  = 12
    T] = YH + Yi2 + Y]3 = 3.612
    T2 = Y2i + Y22 + Y23 « 3.333
    ^3 = Y31 + Y32 + Y33 - 2.605
    T4 = Y41 + Y42 + Y43 - 1.768

    G  = T] + T2 + T3 + T4 =  11.318

          p
    SSB = £ T-j2/ni - Q2/N
                      - (11.318)2  s
              n-i
    SST =
                 - (11.318)2
                      12
               .002
    SSW = SST - SSB - 1.002 - 0.681 - 0.321

    SB2 - SSB/p-1 - 0.681/4-1 - 0.227

    SW2 = SSW/N-p = 0.321/12-4 - 0.040

14.2.8.3  Summarize these calculations in the ANUVA table  (Table  8)


           TABLE 8.  ANOVA  TABLE FOR  DUNNETTS  PROCEDURE  EXAMPLE
Source
Between
Within
df
3
8
Sum of Squares
(SS)
0.681
0.321
Mean Square(MS)
(SS/df)
0.227
0.040
    Total
11
.002
                                    149.

-------
 14.2.8.4  To perform the individual comparisons, calculate the t
 statistic for each concentration, and control combination as follows:
 Where Yi
       ni
                                SWV HAii) + (1/ni)


          * mean proportion surviving for effluent concentration i
          = mean proportion surviving for the control
          = square root of within mean sqaure
          = number of replicates for control
          = number of replicates for concentration i.
 14.2,8.5   Table  9  includes  the  calculated  t  values  for  each
 concentration  and control  combination.   In  this  example,  comparing  the
      concentration with  the control  the  calculation is  as  follows:
1.
                             (  1.204  -  1.111  )


                        [ 0.20  v/	1173'} +  (1/3)   ]
                                                 = 0.570
                        TABLE 9.  CALCULATED T-VALUES
           Effluent Concentration^)
1.0
3.2
10.0
2
3
4
0.570
2.058
3.766
14.2.8.6  Since the purpose of this test is to detect a significant
reduction in survival, a (one-sided) test is appropriate.  The critical
value for this one-sided test is found in Table 5, Appendix C.  For an
overall alpha level of 0.05, eight degrees of freedom for error and three
concentrations (excluding the control) the critical value is 2.42.  The
mean proportion surviving for concentration "i" is considered
significantly less than the mean proportion surviving for the control if
ti is greater than the critical  value.  Therefore, only the 10.0%
concentration has a significantly lower mean proportion surviving than
the control.  Hence the NOEC is  3.2% and the LOEC is 10.0%.

-------
 14.2.8.7  To quantify the sensitivity of the test,  the minimum significant
 difference (MSD)  that can be detected statistically may be calculated.

                    MSD =  d SWV  (1/ni)  + (1/n)

 Where   d  =  the critical  value for  the  Dunnett's  procedure
        SK =  the square root of the  within  mean  square
        n  =  the common number of  replicates  at  each  concentration
             (this  assumes equal replication  at  each  concentration
        n-f =  the number of replicates  in  the  control.

 14.2.8.8  In  this  example:

  -  ;,-               MSD -  2.42 (0.20) /TT/S) +  (1/3F
                        =  2.42 (0.20H0.817)
                        =  0.395

 14.2.8.9   The MSD  (0.395)  is in transformed  units.   To determine the MSD in
 terms of  percent survival,  carry out  the following conversion.

         1. Subtract the MSD from the  transformed control mean*

                              1.204  -  0.395 = 0.809

        2. Obtain the  untransformed values for the control mean and the
           difference  calculated in 4.10.1.

                            [Sine (1.204) ]2 * 0.871
                            [Sine (0.809) ]2 = 0.524

        3. The untransformed MSD (MSDU) is determined by subtracting the
           untransformed values  from 4.10*2.

                          MSDU'= 0.871 - 0.524 = 0.347

 14.2.8.10  Therefore,  for this set of data, the minimum difference in mean
proportion surviving between the  control and any effluent concentration  that
can be detected as statistically significant is 0.347.

 14.2.8.11  This represents a 40%  decrease in survival from the control.
                                      151

-------
 :;;;;;

14.2.9  Probit Analysis    ,J
          dff* use?.r°r the probit analysis is summarized in
           r,re,r^a«^s "s
                                         «-
                   i:

  TABLE 10.  DATA FOR PROBIT ANALYSIS
Number Dead
Number Exposed
               6
              45
       9
       45
             .Effluent Concentration (%)
Control  1.0    3,2   10.0   32.0   100.0
19
45
30
45
45
45
45
45
             152

-------
 TABLE  11.   OUTPUT FROM  EPA PROBIT ANALYSIS PROGRAM.  Version  1.3
             USED FOR CALCULATING EC  VALUES
Probit  Analysis of Inland Silverside Larval  Survival  Data
    Cone.

   control
    1.0000
    3.2000
   10.0000
   32.0000
  100.0000
 Number
Exposed

    45
    45
    45
    45
    45
    45
Number
Resp.

    6
    9
   19
   30
   45
   45
Proportion
Responding

  0.1333
  0.2000
  0.4222
  0.6667
  1.0000
  1.0000
 Adjusted
Proportion
Responding

  0,0000
  0.0483
  0.3126
  0.6034
  1.0000
  1.0000
Predicted
Proportion
Responding

  0.1594
  0.0262
  0.2479
  0.7094
  0.9649
  0.9988
Chi - Square Heterogeneity =
                    4.026
Mu
Sigma

Parainet er

I nt erirept
Slope
Spoutaneous
Response Rate
    0.778527
    0.401388

    Estimate
    std. Err.
             95% confidence Limits
3.060416
2.491352
0.159420
0.445523
0.449330
0.044587
( 2.187191,
{ 1.610665,
( 0.072030,
3.933641)
3. 372039)
0.246810)
      Estimated F.c Values and Confidence Limits
Poi nt
       Cone.

        0.6995
        1.3131
        1.8370
        2.3042
        6.0052
       15.6504
       19.6312
       27.4644
       51,5557
                                      Lower       Upper
                                        Confidence Limits
                             1.4644
                             2.3671
                             3.0701
                             3.6688
                             8.2460
                            24.5887
                            33.5876
                            54.3685
                           138.5833
                                     0. J575
                                     0.4109
                                     0.6825
                                     0.9587
                                     3.8086
                                    11.4046
                                    14.0149
                                    18.6517
                                    30.8662
                                    153

-------
        PLOT OF ADJUSTED PROBITS AND PREDICTED REGRESSION LINE
Probit
   10+
    9 +
    7 +
    6 +
    5 +
    4 +
    3 +
    2 +
        +
      EC01
EC10
EC25
                  EC50      EC75     EC90
                                                   EC99
    figure  4.  Plot of adjusted  probits and predicted  regression line,
                                     154

-------
14.3 ANALYSIS OF GROWTH DATA

14.3.1  Formal statistical analysis of the growth data is outlined in
Figure 5.  The response used in the statistical analysis is mean weight per
replicate.  Concentrations above the NOEC for survival are excluded from the
growth analysis.

14.3.2  The statistical analysis consists of a parametric test, Dunnett's
Procedure, and a non-parametric test. Steel's Many-one Rank Test. The
underlying assumptions of the Dunnett's Procedure, normality and homogeneity
of variance, are formally tested.  The test for normality is the
Shapiro-Wilks Test and Bartlett's Test is used to test for homogeneity of
variance.  If either of these tests fail, the non-parametric test, Steel's
Many-one Rank Test, is used to determine the NOEC and LOEC end points.  If
the assumptions of Dunnett's Procedure are met, the end points are
determined by the parametric test,

14.3.3  Additionally, if unequal numbers of replicates occur among the
concentration levels tested there are parametric and non-parametric
alternative analyses.  The parametric analysis is the Bonferroni t-test. The
Wilcoxon Rank Sum Test with the Bonferroni adjustment is the non-parametric
alternative.  For detailed information on the Bonferroni adjustment, see the
Appendix.

14.3.4  The data, mean and standard deviation of the growth observations at
each concentration including the control are listed in Table 12.  A plot of
the data in Table 12 is provided in Figure 6.  Since there was no survival
in the 32% and  100% concentrations, these are not considered in the growth
analysis.  Additionally, since there is significant mortality in the 10%
effluent concentration, its effect on growth is not considered.
                   TABLE 12.  INLAND SILVERSIDE GROWTH DATA
Replicate    Control
                                   Effluent Concentration (%)
1.0
3.2
10.0   32.0   100.0
A
B
C
Mean (7,-)
Si2
i
0.939
0.976
0.975
0.963
0.0004
1
0.996
1.152
1.066
1.071
0.006
2
0.903
0.864
1.197
0.988
0.033
3
0.491 -
0.589 -
1.131 -
0.737 -
0.119 -
4 5
_
-
."
-
-
6
                                     155

-------
             STATISTICAL  ANALYSIS OF INLAND SILVERSIDE  LARVAL

                           SURVIVAL AND  GROWTH TEST
                                 GROWTH  DATA

               (EXCLUDING CONCENTRATIONS" ABOVE NOEC  FOR SURVIVAL)
                                                             DISTRIBUTION
                   NORMAL DISTRIBUTION
              HETEROGENEOUS
                VARIANCE
                                BARTLE.TT'S TEST
      HOMOGENEOUS VARIANCE
                    EQUAL NUMBER OF

                      REPLICATES?
EQUAL NUMBER OF

  REPLICATES?
          WILCOXON RANK SUM
              TEST WITH
        BONFERRONI ADJUSTMENT
                                STEEL'S MANY-ONE
                               ENDPOINT ESTIMATES
                                    NOEC, LOEC
Figure 5.  Flow chart for statistical  analysis of Menldia growth data.


                                 156

-------
                                             i
                                                     ttJ
                                                     0>
                                                     s-
                                                     o
                                                     
.«'. 1* ....,.,,.	



ddddddddddoddddddddd
 (OH) 1HOGM NV3W
                                       O
                                       d
157

-------
    14.3.5  Test for Normality
                                          a
           ation from each observation  ,'n  that ™  observations within a
   observations are summarized in  Table  ?3     conce^ration.   The  centered
            TABLE  13.  CENTERED OBSERVATIONS FOR SHAPIRO-
                                                       WILKS EXAMPLE
                         0.024
                         0.013
                         0.012
                                         0.075
                                         0.08T
                                         0.005
-0.085
-0.124
 0.209
 H.3.5.2  Calculate the denominator,  D, of the test Stat1st1c

                     D = Z (Xf  - x)2
                        7=1
    Where X7- = the ith  centered observation

          5  == ffi ss^is s
For this  set of data:
                                n B  g
                                     1(0.002) = 0.000
                                D -  0.0794

'4.3.5.3  Order th. centered observations fro, s.allest  to  ,.P9est.

                             -  ... .X(n)


                             ob""«*"'"-  Th.,e ordered observations
          {„'
                                 158

-------
   TABLE  14.
               ORDERED CENTERED OBSERVATIONS FOR SHAPIRO-UILKS EXAMPLE
                        -0.124
                        -0.085
                        -0.075
                        -0.024
                        -0.005
                                                6
                                                7
                                                8
                                                9
0.012
0.013
0.081
0.209
                                  ' *
 n/2.   For the d^th
 listed in Table 15.
                                                                    n,
                                                                 are
14.3.5.5  Compute the test statistic,  W,  as  follows:

                       k
                                      - xd'Jj  ]2
               W = 1 [,r ai-

The differences x(n-i+l) _ x(j)
of data:
                               are  H t d  .
                                    nstea  in Table 15.  For this set
                         0.0794
                                  .2707)2 = n
                                  *t/u/;    u-
  TABLE 15.   COEFFICIENTS AND DIFFERENCES FOR SHAPRIO-WILKS EXAMPLE

                                    	

                         x(n-i+1)  -  x(i)

                                        	' '      	—

                                                 X(9) -
1
2
3
4
0.5888
0.3244
0.1976
0.0947
0.333
0.166
0.088
0.036
                                                  X<7)
                                                  X(6)
                                                        X(3)
                                                        X(4)
                               the
                                  159

-------
 14.3.6  Test  for Homogeneity  of  Variance

 14.3.6.1  The test used to examine whether the  variation  in  mean  dry
 weight  is the same across all effluent concentrations  including the
 control, is Bartlett's Test {Snedecor and Cochran,  1980).  The test
 statistic is as follows:

            R —
                [ ( I Vi) In 52 - I Vi In
     Where V-j =   degrees of freedom for each effluent concen-
                  tration and control, V, = (nj - 1)
                  number of levels of effluent concentration
                  including the control
          C  =  1 +  ( 3{p-l))-l  [ X  1/V-i  -  ( I  v-j)-l  ]
                                 1*1         i=l

          In =  loge

          i  =  1, 2, ..., p where p is the number of concentrations
                            including the control
          n-j = the number of replicates for concentration  i.

14,3.6.2  For the data in this example,  (See Table 12) all effluent
concentrations including the control have the same number of replicates
(rif = 3 for all i).  Thus, V, = 2 for all i.

14.3.6.3  Bartlett's statistic is therefore:


       B =  [{6)ln(0.0132) - 2 I ln(S1)2]/i.25
                              1-1

         -  [6(-4.3275) - 2(ln(0.0004)+ln{0.0061)+ln(0.0331))]/1.25

         -  t-25.965 -  {-32.664)J/1.25

         «  5.359
                                    160

-------
appropriate critical value for this test
with 2 degrees of freedom, is sfz      '
cnt,c.T..v.,u. of 9.270.  concede

14.3.7  Dunnett's Procedure
                                              B -
                                                                - 1 "9"",
                                                      -  .The)"efore, the

                                                              7eVe] °f °'0]
     Source
    Total
                            TABLE 16.   ANOVA TABLE
               N - 1
                             Sum of Squares
                                  (SS)
Between
Within
P - 1
N - p
SSB
SSW
                                 SST
                                                 Mean  Square(MS)
                                                     (SS/dfJ
                                                   SB  =  SSB/(p-l)
                                                   2

                                                        SSW/{N-p)
Where:
            nf  =  number  of  observations  in concentraion*!  P
              * I T7-2/ni - Q2/N
                                       Between Sum of Squares

          SST » z   zt 2 . G2/N
               i=l j=1

          SSW = SST -  SSB
                                       Total  Sum of Squares


                                       Within Sum of Squares
                the grand total of all sample observations, G = j TI-

                                                       for
        iJ
                hP mpL°HServation for Concentration »i» (represents
               the mean dry weight of the fish for effuent
               concentration f in test chamber jj   6TTIuent
                                 161

-------
    14.3.7.2  For the data in this
                            example:
 T3  -

 G  =


SSB =
                  tf   f Y]3 = 0.939 + 0.
                  Y22  + Y23 - C
                         = 2.890

 -  0.903  +  0.864  +i:?97==23;92644
                T2 + T3 = 9.068
            1_(27.467)  -
                   ii-068)2
                      9
                                      0.019
               n.
         = 9-235 - (9^68)2  , ^



         -  SST  -  SSB » 0.098  -  0.0)9 = 0.079

    SB2 - SSB/p-1 , 0.019/3-1  . 0.009

    SW2 = SSW/N-p = 0.079/9-3 = 0.013


'4-3.7.3   s«.rfze these  calculations  in  the  ANOVA  fbl.  (Table  ,7)
          TABLE 17.  ANOVA TABLE FOR DUNNETT'S PROCEDURE EXAMPLE
   Source
           df
Sum of Squares
     (SS)
                                                 Mean Square(MS)
                                                     (SS/df)
   Between         2

   Within

-------
    "ere f;
           W
                                        < Yl - Yl )
                                       —	—

                                   Sh/N/TlA)]) + (]/nf
                    g

             - square root of within mean  sqaure
             - number of replicates  for control
             - number of replicates  for' concentration i
   1.0*  concentration
                           t  =
                        TABLE 18.   CALCULATED T-VALUES
                        •••

            Effluent Concentration
                          .-             t,        .»".„*
      ca   value for this one-sided test it fn  , ?PProPrlate.  The
 For  an overall alpha level of 0 05  six rf«  "d ^ Iable 5» APPe"dix C
 two  concentrations (excluding the wntrol)9th^Sr0-/re?doni for error
 The  mean weight for concentration •?"  is  rnnS?rff "J10?1 Value fs 2'34-
 than mean weight for the control  if \   •  conside''ed significantly less
 value.  Therefore,  all  efS concelltMt?n«t-r ^an the Criti"l
have significantly  lower mean  weights than V^f  '" Jhl? example do not
and the  LOEC  for  growth  can not bf calculated       °U   He"Ce the NOEC
calculated,
                                                 statistically may be
                                  163

-------
   Where  d  =
          n  =
          m =
   14.3.7.8 In  this  example:
  can be'
                         = 2.34 (0.114) VTT73TTT
                         "2.34 (O.T14H0.8T65
                         - 0.218

               1S  represe"ts
                                    reduction fn niean .eight fro, the
  15-  PRECISION AND

  15.1  PRECISION
                an^ S S°r'*°E
sulfate as reference^oxicants
Tables 19 and 20.   in  the      '
                                                of the inland ,
                                                     and sodful"
                                                    are  ™1d   in
                                                                 '  the
15.2  ACCURACY

15-2.1   The accuracy of toxicity tests cannot be determined.
                                  164

-------
   TABLE  19. SINGLE LABORATORY PRECISION OF THE INLAND SILVERSIOE
            (MENIDIA BERYLLINA) SURVIVAL AND GROWTH TEST  PERFORMED IN

            =1L *™™:. ™ rAE FROM FISH ^«YN
                              SEAWATER, AND COPPER AS /
                Survival
  Test
             NOEC
            (ug/LJ
             63
            125
            135
            125
            125
                                          Growth
 LOEC
(ug/L)
                       125
                       250
                       250
                       250
                      250
NOEC
(ug/L)
••"— 	 	 • r- • i
125
125
63
125
31
LOEC
(ug/L)
— - i i — I,,
SE
SE
125
SE
63
  Most
Sensitive
End Point
                                      S

                                      G

                                      G
 'Tests  performed by George Morrison  and Elise Torello
 were: 31,  63,  125, 250,and 500 ug/u
3Adults collected  in the field.
                                                   co"ce"trations
5SE = Survival effects.  Growth data  at  these

                                        '
                                165

-------
    ™LE
  Test
                                        ™
             SKATER. USING LARVAE FROM
   luryjvaj
NOEC       L0£c

            (mg/L)
                                            Growth
NOEC
                                                 (mg/L)
                         Most
                      Sensitive
                      End Point
*t = Survival Effprtc   r
concentrations were disreaaTL/?*'  at  these  ^^'cant
reduction in sUrviVa?.Sre9arded  because there was a significant
                                166

-------
 o  o
 s- •«-
 c
     O
 t/i  c:
     13
 S_  Jfl     gj
 oj  >
          03   —
          Q.    03

          W   u.
S- 0
o
OJ
s_
3
cn
         (D   -^
        Q   UJ
              W     T
              o.   ~
              >-   •£
                                       O
                                      o
" I

2?
                                                             -^ "*
                                                             * o

                                                                                                                                         si:
                                                                                                                           i ^
                                                                                   C/)
                                                                                   t—

                                                                                   LLJ

                                                                                   5


                                                                                   o
                                                                                            aj
                                                                                                                                                          OJ 00

                                                                                                                                                         IET (Tl
     O




     a>

 nj


-------
-o
 (U
               .  .2
            c/)  u.
              »   ^
              •-   ^z
              n   *•-
                   a.   =



t-
u
a.
hU






H
<
t.
a
u






1-
<
t
C
L





z
o
cc
t-
UJ
O
z
•0 O
*•
tt
Effluent 1
r*
(O
^
^
m
«
-
o
r-
(D
if)

	
1
tN
-
O
^
to

I »
J
u "

^
o
r*
to
to
*
5 «
u
£ "
UJ —
CC ^.
o
V
>
<
c
































) J
»-
































! a.
































fjumnvs I
































i 0
0

o
m
2 ?
i —
« c
£2


*<;
3B






*-
K *
u a

a
|f
M


?-=
* o


a
j i
<
a*



is
• 5
t S
* *



























_
^
UJ
O
o

^
































h
•
































U
































-^ 3 1
J < •-

O
il
li


*&
is






Ul
< >

O
?'

il


*s
h


c
I/
t- +
J 1
0 '



ul
ff!


























z
—
—
. t-
LX
. z
UJ
CJ
o
o

u
t „!
5 >B
1 -
































TEMP
m
SALINITY
































2

Q
X -^
s ?
i!


5*
* a






Is
« a

o
«
i- •
i
15 5*
Ji
si



-------
             Figure 8. Data forms for inland silverside larval survival
                       and growth test.   Dry weights of larvaeJ
[Test Dates:
                      Species:
        Pan
         #
Cone,
  &
 Rep.
Initial
 Wt.
Final
 Wt.
(mg)
Diff.
(mg)
  #
Larvae
Av. Wt./
Larvae
 (mg)
    'Adapted from: M. A. Heber, M. M. Hughes, S. C. Schimmel, and
        D. Bengtson, 1987
                                        169

-------
 Tost Dates:
 Figure 9.  Data forms for  inland  silverside larval
      survival and growth  test.   Summary of test
      results. '
	      Species:
 Effluent Tested:
TREATMENT
# LIVE
LARVAE
SURVIVAL
{%)
MEAN DRY WT./
LARVAE fmg)
±S.D.
SIGNIF. DIFF.
FROM CONTROL
(o)
MEAN
TEMPERATURE
(oC)
+ S.D.
MEAN SALINITY
000
±S.D.
AV. DISSOLVED
OXYGEN
(mg.t) +S.D



































••MM^M^H






COMMENTS;
           ^Adapted from: Heber, M.
             and D. Bengston, 1987.
                 A., M. M. Hughes,  S.  C.  Sctiimnel,
                                       70

-------
                                   SECTION 14

                                 TEST METHOD1'2

         MYSID  (MYSIDOPSIS BAHl/\) SURVIVAL,  GROWTH, AND FECUNDITY TEST
                                  METHOD  1007

1.  SCOPE AND APPLICATION

1.1  This method estimates the chronic toxicity of effluents and receiving
waters to the estuarine mysid, Mysidopsis bahia, during a seven-day,  <
static-renewal exposure.  The effects include the synergistic, antagonistic,
and additive effects of all the chemical, physical, and additive components
which adversely affect the physiological and biochemical functions of the test
organisms.

1,2  Detection limits of the toxicity of an effluent or pure substance are
organism dependent.

1 3  Single or multiple excursions in toxicity may not be detected using 24-h
composite samples.  Also, because of the long sample collection period
involved in composite sampling and because the test chambers are not sealed,
highly volatile and highly degradable toxicants in the source may not be
detected in the test.

1.4  This method  should be restricted to use by, or under the supervision^of,
professionals experienced  in aquatic toxicity testing.  Specialized training
is required to determine the sex of the maturing mysids and the presence of
eggs in the oviducts of the females.

2.  SUMMARY OF METHOD

2 1  This rapid-chronic test consists of  an  exposure of 7-day old Mysidopsis
bahia juveniles to  different concentrations  of  effluent, or to receiving water
Tna static  system, during the period of  egg development. The test  end points
are survival, growth  (measured as dry weight),  and fecundity  {measured as  the
percentage  of females with eggs  in the  oviduct  and/or  brood pouch).

3.  DEFINITIONS

     (Reserved for addition of  terms  at  a  later  date).

4.  INTERFERENCES

4.1   Toxic  substances  may be  introduced by contaminants  in  dilution water,
 glassware,  sample hardware,  and  testing equipment (see Section  5, Facilities
 and  Equipment).
 ^The  format used  for  this method was taken from Kopp,  1983.
 2This  method was  adapted from Lussier,  Kuhn, and Sewall.  1987,  Environmental  ,
 Research Laboratory,  U. S. Environmental  Protection  Agency,  Narragansett,
 Rhode Island.                               r;

-------
4.2  Improper effluent sampling and handling may adversely affect test
results (see Section 8, Effluent and Receiving Water Sampling and Sample
Handling).

4.3  The test results can be confounded by (1) the presence of pathogenic
and/or predatory organisms in the dilution water and effluent, (2) the
condition of the brood stock from which the test animals were taken, (3) the
amount and type of natural food in the effluent or dilution water, (4)
nutritional value of the Artemia nauplii fed during the test, and (5) the
quantity of Artemia nauplii or other food added during the test,  which may
sequester metals and other toxic substances, and lower the DO.

5.  SAFETY

5.1  See Section 3, Health and Safety.

6.  APPARATUS AND EQUIPMENT

6.1  Facilities for holding and acclimating test organisms.

6.2  Brine shrimp culture unit -- see paragraph 7.12 below.

6.3  Mysid culture unit — see Paragraph 7 below.  This test requires a
minimum of 240 7-day old {juvenile} mysids.  It is preferable to  obtain the
test organisms from an inhouse culture unit.  If it is not feasible to
culture mysids inhouse, juveniles can be obtained from other sources, if
shipped in well oxygenated saline water in insulated containers.

6.4  Samplers -- automatic sampler, preferrably with sample cooling
capability, that can collect a 24-h composite sample of 5 L.

6.5  Environmental chamber or equivalent facility with temperature control
(26 + IOC).

6.6  Water purification system -- Millipore Super-Q, deionized water or
equivalent.

6.7  Balance — capable of accurately weighing to 0.000001 g.

6.8  Reference weights, Class S — for checking performance of balance.
Reference weights should bracket the expected weights of the weighing boats
and weighing boats plus organisms.

6.9  Drying oven -- 105°C, for drying organisms.

6.10  Desiccator -- for holding dried organisms.

6.11  Air pump — for  supplying air.

6.12  Air  lines, and air stones -- for aerating cultures, brood chambers,   -,'
and holding tanks, and supplying air to test solutions with low DO.
                                      172

-------
 6.13   pH  and  DO  meters  --  for routine  physical  and  chemical measurements.
 Unless  the  test  is being conducted  to  specifically  measure the  effect  of one
 of  the  above  parameters, a  portable, field-grade  instrument is  acceptable.

 6.14  Tray  --  for test  vessels; approximately 90  X  48 cm to hold 56 vessels.

 6.15  Standard or micro-Winkler apparatus -- for  determining DO and checking
 DO  meters

 6.16  Dissecting microscope  (350-400X magnification) -- for examining
 organisms in  the test vessels to determine their  sex and to check for  the
 presence  of eggs in the oviducts of the females.

 6.17  Light box -- for  illuminating organisms during examination.

 6.18  Refractometer or other method-- for determining salinity.

 6.19  Thermometers, glass or electronic, laboratory grade -- for measuring
 water temperatures.

 6.20  Thermometers, bulb-thermograph or electronic-chart type -- for
 continuously recording temperature.

 6.21  Thermometer,  National Bureau of Standards Certified (see USEPA METHOD
 170.1, USEPA,  1979} -- to calibrate laboratory thermometers.

 6.22  Test vessels -- 200 mi borosilicate glass beakers or 8 oz disposable
 plastic cups  (manufactured by Falcon Division of Becton, Dickinson Co., 1950
 Williams Dr., Oxnard, CA 93030}  or other similar containers.   Cups must be
 rinsed thorougnly in distilled or deionized water and then pre-soaked
 (conditioned) overnight in dilution water before use.  Forty-eight (48) test
 vessels are required for each test (eight replicates at each of five
 effluent concentrations and a control).

 6.23  Beakers or flasks -- six,  borosilicate glass or non-toxic plasticware,
 2000 ml for making test solutions.

 6.24  Wash bottles  -- for deionized water,  for washing organisms from
 containers and for  rinsing small glassware and instrument electrodes and
 probes.

 6.25  Volumetric flasks and graduated cylinders -- Class A,  borosilicate
 glass or non-toxic  plastic labware, 50-2000 ml for making test solutions.

6.26  Separatory funnels, 2-1 -- Two-four for culturing Artemia.

6.27  Pipets, volumetric — Class A, 1-100 nt.

6.28  Pipets, automatic -- adjustable,   1-100 ml.

6.29  Pipets, serological -- 1-10 ml,  graduated.
                                      173

-------
6.30  Pipet bulbs and fillers — PROPIPETR, or equivalent.

6.31  Droppers, and glass tubing with fire polished edges,  4mm ID — for
transferring organisms.

6.32  Forceps — for transferring organisms to weighing boats.

6.33  NITEXR mesh sieves {150 urn and 1000 urn)  — for concentrating
organisms.

6.34  Depression glass slides or depression spot plates --  two, for
observing organisms.

7.  REAGENTS AND CONSUMABLE MATERIALS

7.1  Sample containers -- for sample shipment  and storage (see Section 8,
Effluent and Receiving Water Sampling and Sample Handling).

7.2  Data sheets {one set per test) — for data recording {Figures 14, 15,
and 16).

7.3  Tape, colored and markers, water-proof — for labelling and marking
test chambers, containers, etc.

7.4  Weighing boats, aluminum  — to determine the dry weight of organisms.

7.5  Buffers, pH 4, 7, and 10 (or as per instructions of instrument
manufacturer) -- for standards and calibration check (see USEPA Method
150.1, USEPA, 1979).

7.6  Membranes and filling solutions for dissolved oxygen probe (see USEPA
Method 360.1, USEPA, 1979), or reagents for modified Winkler analysis.

7.7.  Laboratory quality assurance samples and standards for the above
methods.

7.8  Reference toxicant solutions (see Section 4» Quality Assurance).

7.9  Reagent water -- defined as distilled or deionized water that does not
contain substances which are toxic to the test organisms {see paragraph 6.6
above).

7.10  Effluent, surface water, and dilution water — see Section 7, Dilution
Water, and Section 8, Effluent and Surface Water Sampling and Sample
Handling.  Dilution water containing organisms that might prey upon or
otherwise interfere with the test organisms should be filtered through a
fine mesh net  {with 150 urn or smaller openings).

7.10.1  Saline test and dilution water — The salinity of the test water
must be in the range of 20 %o to 30 o/oo.  The salinity should vary by
                                      174

-------
no more than + 2 °/oo among the chambers on a given day.  If effluent and
receiving water tests are conducted concurrently, the salinities of these
tests should be similar.

7.10.1.1 The overwhelming majority of industrial and sewage treatment
effluents entering marine and estuarine systems contain little or no
measurable salts.  Exposure of mysids to these effluents will require
adjustments in the salinity of the test solutions.  It is important to
maintain a constant salinity across all treatments.   In addition, it may be
desirable to match the test salinity with that of the receiving water.
Although artificial sea salts have been shown to be acceptable for
life-cycle toxicity tests with mysids (Home, et al, 1983; ASTM, 1986), the
use of artificial sea salts in this test is not recommended at this time.
Hypersaline brine derived from natural seawater should be used to adjust
salinities.  However, it should be noted that with (100 °/oo) hypersaline
brine, the maximum concentration of effluent that can be tested is 80%
effluent at 20 °/oo salinity and 70% effluent at 70 °/oo salinity.

7.10.1.2  Hypersaline brine -(HSB) has several advantages that make it
desirable for use in toxicity testing.  It can be made from any high
quality, filtered seawater by evaporation, and can be added to the effluent
or to deionized water to increase the salinity.  Brine derived from natural
seawater contains the necessary trace metals, biogenic colloids, and some of
the microbial components necessary for adequate growth,  survival, and/or
reproduction of marine and estuarine organisms, and  may be stored for
prolonged periods without any apparent degradation.

7,10.1.3  The Ideal container for making brine from natural  seawater is one
that (1) has a high surface to volume ratio, (2) is  made of a non-corrosive
material, and (3) is easily cleaned (fiberglass containers are ideal).
Special  care should be used to prevent any toxic materials from coming in
contact  with the seawater being used to generate the brine.   If a heater is
immersed directly into the seawater, ensure that the heater  materials do not
corrode  or leach any substances that would contaminate the brine.  One
successful method used is a thermostatically controlled heat exchanger made
from fiberglass.  For aeration, use only oil-free air compressors to prevent
contamination.

7.10.1.4  Before adding seawater to the brine generator,  thoroughly clean
the generator, aeration supply tube, heater, and any other materials that
will be  in direct contact with the brine.  A good quality biodegradable
detergent should be used, followed by several thorough deionized water
rinses.   High quality (and preferably high salinity) seawater should be
filtered to at least 10 urn before placing into the brine generator.  Water
should be collected on an incoming tide to minimize  the possibility of
contamination.

7.10.1.5  The temperature of the seawater is increased slowly to 40°C.
The water should be aerated to prevent temperature stratification and to
increase water evaporation.  The brine should be checked daily (depending on
the volume being generated) to ensure that the salinity does not exceed
                                      175

-------
100 °/oo and that the temperature does not exceed 4Q°C,   Additional
seawater may be added to the brine to obtain the volume  of brine required.

7.10,1.6  After the required salinity is attained, the HSB should be
filtered a second time through a 10 urn filter and poured directly into
portable containers (20-L cubitainers or polycarbonate water cooler jugs are
suitable).  The containers should be capped and labelled with the date the
brine was generated and its salinity.  Containers of HSB should be stored in
the dark and maintained under room temperature until used.

7.10.1.7  If a source of HSB is available, test solutions can be made by
following the directions below:

7.10.1.8  Thoroughly mix together the deionized water and brine before
mixing in the effluent.  Divide the salinity of the HSB  by the expected test
salinity to determine the proportion of deionized water  to brine.  For
example, if the salinity of the brine is 100 /oo salinity from a HSB of
100 o/oo, 200 ml of brine and 800 ml of deionized water are required.

7.10.1.9  Table 1 illustrates the composition of 3-L test solutions at 20
o/oo if they are made by combining effluent (0 °/oo), deionized water
and HSB of 100 o/oo (only).  The volume (ml) of brine required is
determined by using the amount calculated above.  In this case, 200 ml of
brine  is required for 1 L; therefore, 600 ml would be required for 3 L of
solution.  The volumes of HSB required are constant.  The volumes of
deionized water are determined by subtracting the volumes of effluent and
brine  from the total volume of solution:  3000 ml - ml effluent - ml brine =
ml deionized water.

7.12 .BRINE SHRIMP (ARTEMIA) NAUPLII {see Peltier and Weber, 1985).

7.12.1  Newly hatched Artemia nauplii are used for food for the stock
cultures and test organisms.  Although there are many commercial sources of
brine  shrimp cysts, the Brazilian or Colombian strains are preferred because
the supplies examined have had low concentrations of chemical residues and
produce nauplii of suitably small size.   (One source that has been found to
be acceptable  is Aquarium Products,  180L  Penrod Ct., Glen Burnie, Maryland
21061).  Each new batch of Artemia cysts must be evaluated for size
(Vanhaecke and Sorgeloos, 1980,  and  Vanhaecke et al., 1980) and nutritional
suitability  (see Leger et al., 1985, 1986) against known  suitable reference
cysts  by performing a side-by-side larval growth test using the "new" and
"reference" cysts.  The "reference"  cysts used in the suitability test may
be a previously tested and acceptable batch of cysts, or  may be obtained
from the Quality Assurance Branch, Environmental Monitoring and Support
Laboratory,  Cincinnati, Ohio.  A sample of newly-hatched  Artemia nauplii
from each new  batch of cysts  should  be chemically analyzed.If the,
concentration  of total organic chlorine exceeds 0.15 ug/g wet weight, or  the
                                      176

-------
TABLE 1.  QUANTITIES OF EFFLUENT, DEIONIZED WATER, AND HYPERSALINE BRINE
          (100 o/oo) NEEDED TO PREPARE 1800 ML VOLUMES OF TEST SOLUTION
          WITH A SALINITY OF 20 o/oo.
Effluent
Concentration
'(*)
33
10
3.3
1.0
0.33
Control
Volume of
Effluent
(0 o/oo)
(iML)
600
200
67
22
7.3
—
Volume of
Deionized
Water
(mL)
840
1240
1373
1418
1433
1440
Volume of
Hypersaline
Brine
(mL) j.
360
360
360
360
360
360
              Total
896
7744
2160
                                     177

-------
total concentration of organochlorine pesticides plus PCBs exceeds 0.3 ug/g
wet weight, the Arternla cysts should not be used (For analytical method see
EPA, 1982).

7.1?. 2  Artemia nauplii are obtained as follows:

    1.   Add 1 L of seawater, or an aqueous uniodized salt (NaCl) solution
        prepared with 35 g salt or artificial sea salts per liter, to a 2-L
        separatory funnel, or equivalent.
    2.   Add 10 ml Artemia cysts to the separatory funnel and aerate for 24 h
        at 27°C,  Hatching time varies with incubation temperature and the
        geographic strain of Artemia used.  See Peltier and Weber (1985),
        for details on Artemia culture and quality control.
    3.   After 24 h, cufoTTTFe air supply in the separatory funnel.
        Artemia nauplii are phototactic, and will concentrate at the bottom
        of the "funnel if  it is covered for 5-10 min with a dark cloth or
        paper towel.  Caution: if the concentrated nauplii are  left on the
        bottom much longer than 10 min without aeration, excessive mortality
        will result.
    4.  Drain the nauplii into a funnel fitted with a 150 urn Nitex screen,
        and rinse with seawater or equivalent before use.
    5.  Resuspend the nauplii on the funnel in a small amount of water for
        feeding.

7.12.3  Testing the acceptability of Arteinia, nauplii as food for mysids.

7.12.3.1   The primary criteria for acceptability of each new supply of brine
shrimp 'cysts  is adequate  survival, growth, and reproduction of  the mysids.
The tnysids used to evaluate the acceptability of the brine shrimp nauplii
must be of the  same geographical origin and stage of development  (7 days
old) as those used routinely  in the toxicity tests.  Two 7-day  chrome tests
are  performed side-by-side, each consisting of eight replicate  test vessels
containing five juveniles  (40 organisms per test, total of BO organisms).
The  juveniles  in one  set  of test chambers  is fed reference (acceptable)
nauplii and  the other  set  is  fed nauplii  from the "new" source  of Artemia
cysts.

7.12.3.2   The  feeding  rate  and frequency,  test  vessels, volume  of control
water,  duration of  the  test,  and age of the nauplii at  the start  of the
       should  be the  same  as used for the  routine toxicity  tests.
test,
                                                                    are  only
                                                                     The
7.12.3.3  Results of the brine shrimp nutrition assay,  where there
two treatments,  can be evaluated statistically by use of a t-test.
"new" food is acceptable if there are no statistically significant
differences in the survival, growth, and reproduction of the mysids fed the
two sources of nauplii.
7.13 MYSIDS
               (See Rodgers,  et  al.,  1986,  and
               information on inysid  ecology)
                                              Peltier and Weber, 1985, for
7.13.1  Brood Stock
7.13.1.1
cultures
           To provide an adequate supply of juveniles  for  a  test,  mysid
          should be started at least four weeks before the test animals
are
                                      178

-------
needed.  At  least 200 mysids should be placed  in each culture tank to ensure
that  1500 to 2000 animals will be available by the time preparations for a
test  are initiated.

7.13,1.2  Mysids may be shipped or otherwise transported in polyethylene
bottles or CUBITAINERSR.  Place 50 animals in 700 ml of seawater in a 1-L
shipping container.  To control bacterial growth and prevent DO depletion
during shipment, do not add food.  Before closing the shipping container,
oxygenate the water for 10 min.  The mysids will starve if not fed within
36 h, therefore, they should be shipped so that they are not in transit more
than  24 h.

7.13.1.3  The identification of. the stock culture should be verified using
the key from Heard, Price and Stuck, 1987.  Records of the verification
should be retained along with a few of the preserved specimens.

7.13.1.4  Glass aquaria {120- to 200-L) are recommended for cultures.   Other
types of culture chambers may also be convenient.  Three or more separate
cultures should be maintained to protect against loss of the entire culture
stock in case of accident, low DO, or high nitrite levels,  and to provide
sufficient numbers of juvenile mysids for toxicity tests.   Fill the aquaria
about three-fourths full of seawater.   A flow-through system is recommended
if sufficient natural seawater is available,  but a closed,  recirciflating or
static renewal  system may be used if proper water conditioning is provided
and care is exercised to keep the pH above 7,8 and nitrite  levels below
0.05 mg/L.

7.13.1.5  Standard aquarium undergravel filters should be  used with either
the flow-through or recirculating system to provide aeration and a  current
conducive to feeding (Gentile et al.»  1983).  The undergravel filter is
covered with a  prewashed,  coarse T^-5  mm)  dolomite substrate,  2.5 cm deep
for flow-through cultures  or 10 cm deep for recirculating  cultures.

7.13.1.6  The recirculating culture system is  conditioned  as follows:

    1. After the dolomite  has been added,  the  filter  is  attached to the  air
       supply and operated for  24 h.
    2. Approximately. 4 L of seed water obtained from  a successfully
       operating culture is added to the  culture chamber.
    3. The  nitrite level is checked daily with  an aquarium  test kit or with
       EPA  Method 354.1.  (USEPA,  1979b).
    4. Add  about 30 ml of  concentrated Artemia  nauplii  every other  day  until
       the  nitrite level reaches at least  2.0 mg/L. The  nitrite will
       continue  to rise for several  days without adding  more Artemia and
       will  then slowly decrease to less  than 0.05  mg/L.
    5. After the nitrite level  falls  below 0.05  mg/L,  add another 30 mL  of
       Artemia  nauplii  concentrate and check  the nitrite concentration every
       day,
    6. Continue  this cycle until  the addition of Artemia nauplii  does not
       cause a rise in the nitrite concentration.   The culture  chamber  is
       then  conditioned  and is  ready to receive  mysids.
    7. Add  only  a few (5-20)  mysids  at first, to  determine  if  conditions are
       favorable. If these mysids  are  still doing well  after a  week, several
       hundred more  can  be added.
                                      179  ;;•

-------
7.13.1.7  It  is important to add enough food to keep the adult animals from
cannibalizing the young, but not so much that the DO is depleted or that
there is a build up of toxic concentrations of ammonia and nitrite.  Just
enough newly-hatched Artemia nauplii are fed twice a day so that each
feeding is consumed before the next feeding.

7.13.1.8  Natural sea water is recommended as the culture medium, but HSB
may be used to make up the culture water if natural sea water is not
available.

7.13.1.9  Mysidopsis bahla should be cultured at a temperature of
25 + 2°C.  No water temperature control equipment is needed if the ambient
laboratory temperature remains in the recommended range, and if there are no
frequent,  rapid, large temperature excursions in the culture room.

7.13.1.10   The salinity should be maintained at 30 +_ 2 °/oo, or at a lower
salinity (but not less than 20 °/oo) if most of the tests will be
conducted  at a lower salinity.

7.13.1.11   Day/night cycles prevailing in most laboratories will provide
adequate illumination for normal growth and reproduction. A 16-h/8-h
day/night  cycle in which the light is gradually increased and decreased to
simulate dawn and dusk conditions, is recommended.

7.13.1.12   Mysids cannot survive DOs below 5 mg/L for extended periods. The
airlift used in most undergravel filters will usually .provide sufficient
DO.  If the DO drops below 60% saturation (4.8 mg/L at  25°C and 30 ppt
salinity;  see Section 8), additional aeration should be provided.  Measure
the DO in  the cultures daily during the first week and  then at least weekly
thereafter.

7.13.1.13   Suspend a clear glass or plastic panel over  the cultures, or use
some other means of excluding dust and dirt, but leave  enough space between
the covers and culture tanks to allow circulation of air over the cultures.

7.13.1.14   If hydroids or worms appear in the cultures, remove the mysids
and clean  the chambers thoroughly, using soap and hot water.  Rinse once
with acid  (10% HC1) and three times with distilled or deionized water.
Mysids with attached hydroids should be discarded.  Those without hydroids
should be  transferred by hand pipeting into three changes of clean seawater
before returning them to the cleaned culture chamber.  To guard against
predators, natural sea water should be filtered through a net with 30 urn
mesh openings before entering the culture vessels.

7.13.1.15   Mysidopsis bahia are very sensitive to low pH and sudden changes
in temperature.  Care should be taken to maintain the pH at 8.0 + 0.3, and
to limit rapid changes in water temperature to less than 3°C.
                                      180

-------
  7.13.1.16   Mysids  should  be  handled  carefully and as  little as possible  so
  that  they  are  not  unnecessarily  stressed or  injured.  They should be
  transferred between  culture  chambers with  long handled cups with netted
  bottoms.   Animals  should  be  transferred to the test vessels with a  large
  bore  pipette  (4-mm), taking  care to release  the animals under the surface of
  the water.   Discard  any mysids that are injured during handling,

  7.13.1.17   Culture Maintenance

  7.13.1.17.1  Cultures in  closed, recirculating systems are fed twice a day.
  If no nauplii are present in the culture chamber after four hours, the
  amount of  food should be  increased slightly.   In flow-through systems,      ;
  excess food can be a problem by promoting bacterial  growth and low dissolved
  oxygen.

  7.13.1.17.2  Careful culture maintenance is essential.  The organisms should
 not be allowed to become too crowded.  The  cultures  should be cropped as
 often as necessary to maintain a density of about  20 mysids per  liter.  At
 this density, at least  70% of the females  should have eggs in  their  brood
 pouch.  If  they do not,  the cultures  are probably  under  stress,  and  the
 cause should be found and  corrected.   If the  cause cannot  be  found,  it may
 be necessary to re-start the cultures with  a  clean culture chamber,  a  new
 batch of culture water,  and clean gravel.

 7.13.1.17.3  In closed,  recirculating  systems, about  one third of  the
 culture water should  be  replaced  with  newly prepared  seawater every  week.
 Before siphoning the  old media  from the  culture, it  is recommended that the
 sides  of the vessel  be scraped  and  the gravel carefully turned over  to
 prevent excessive build  up of algal growth.   Twice a year  the mysids should
 be removed  from the recirculating cultures, the gravel rinsed in clean
 seawater, the sides of the chamber washed with clean seawater, and the
 gravel  and  animals  returned to  the  culture  vessel with newly conditioned
 seawater. No detergent should be  used, and  care should be  taken not  to  rinse
 all  the  bacteria  from the  gravel.

 7.13.2   Test Organisms

 7.13.2.1  The test  is begun with 7-day old juveniles.  To have the test
 animals available and acclimated to test conditions at the start of the
 test, they  must be obtained from the stock culture eight days in advance of
 the test.   Whenever possible, brood stock should be obtained from cultures
 having similar salinity, temperature,  light regime, etc., as are to be used
 in the toxicity test.

 7.13.2.2  Eight days before the test is to start,  sufficient gravid females
 are placed  in brood chambers.  Assuming that 240 juveniles  are needed for
 each test,  approximately half this number (120)  of  gravid females should be
 transferred to brood chambers.  The mysids are removed from the culture tank
with a net or netted cup and placed in 20-cm diameter finger bowls.  The
gravid females are transferred from the finger bowls  to the brood chambers
with a large-bore pipette or,  alternatively, are  transferred by pouring the
contents of the finger bowls into the  water  in the  brood  chambers.
                                      181

-------
 7.13.2.3  The mysid juveniles may be collected for the toxicity tests by   .,
 transferring gravid females from the stock cultures to netted {1000 urn)
 flow-through containers (Figure 1}  held within 4-L glass,  wide-mouth
 separatory funnels.   Newly released juveniles can pass through the netting,
 whereas  the females are retained.   The gravid females are  fed newly hatched
 Artemia  nauplii,  and are held overnight to permit the release of young.  The
 juvenile mysids  are collected by opening the stopcock on the  funnel and
 collecting them  in  another container from which  they are transferred to
 holding  tanks using a wide bore {4  mm ID)  pipette.   The brood chambers
 usually  require  aeration to maintain sufficient  DO and to  keep the food in
 suspension.

 7.13.2.4  The temperature in the brood chamber should be maintained at the
 upper  acceptable  culture limit  (26  - 27°C),  or 1°C  higher  than the
 cultures,  to  encourage  faster brood  release.   At  this temperature,
 sufficient juveniles  should be  produced  for  the  test.

 7.13.2.5  The newly  released juveniles (age  =  0  days)  are  transferred  to
 20-L glass aquaria  (holding vessels)  which are gently aerated.  Smaller
 holding  vessels may  be  used,  but the  density  of  organisms  should  not exceed
 20 mysids  per liter.  The  test  animals are held  in  the  holding  vessel  for
 six days prior to initiation  of the  test.  The holding  medium is  renewed
 every  other day.

 7.13.2.6   During the  holding  period,  the mysids are acclimated  to  the
 salinity at which the test  will be conducted,  unless  already  at that
 salinity.  The salinity  should  be changed no more than  2 °/oo  per  24 h  to
minimize stress on the juveniles.

 7.13.2.7   The temperature during the holding period is  critical to  mysid
development,  and must be maintained at 26 - 27°C. If the temperature
cannot be maintained  in this range,  it is advisable to  hold the juveniles  an
additional day before beginning the test.

7.13.2.8   During the holding period, just enough newly-hatched Artemia
nauplii are fed twice a day  (a total of at least 150 nauplii per mysid per
day) so that some food is constantly present.

7.13.2.9   If the test is to be performed in the field, the  juvenile mysids
should be gently siphoned into 1-L polyethylene wide-mouth  jars with
screw-cap  lids filled two-thirds full with clean  seawater  from the  holding
tank.   The water in these jars is aerated for 10 min, and  the jars  are
capped and packed in insulated boxes for shipment to the test site.  Food
should not be added to the jars to prevent the development  of excessive
bacterial growth and a reduction in DO.
                                     182

-------
                     INFLOW
                                 OUTFLOW
                             .NETTED
                               CHAMBER


                             .SEPARATORY
                               .FUNNEL   £
                           NETTED
                            CHAMBER
Figure  1. Apparatus (brood chamber) for collection of
         juvenile mysids.  From Lussier, Kuhn, and
         Sewall, 1987.
                       183

-------
 7.13.2.10   Upon  arrival  at  the  test  site  (in  less  than 24 h) the mysids are
 gently poured  from the  jars  into  20-cm diameter  glass culture dishes. The
 jars  are rinsed  with  salt water to dislodge any  mysids that may adhere to
 the sides.   If the water appears  milky, siphon off half of it with a netted
 funnel {to  avoid siphoning  the  mysids} and replace with clean salt water of
 the same salinity and temperature.   If no Artemia  nauplii are present in the
 dishes, feed about 150  nauplii  per mysid.

 8.  SAMPLE  COLLECTION,  PRESERVATION  AND HANDLING

 8.1   See Section 8, Effluent and  Receiving Water Sampling and Sample
 Handling.

!9.  CALIBRATION
t     -i i i •!       '

 9.1   See Section 4, Quality Assurance.

 10.   QUALITY CONTROL

 10.1   See Section  4, Quality Assurance.

 10.2   The reference toxicant recommended for use with the mysid 7-day test
 is copper sulfate.

 11.   TEST PROCEDURE

 11.1   TEST  DESIGN               .;

 11.1.1  The test  consists of at least five effluent concentrations plus a
 site  water  control and a reference water treatment (natural  seawater or
 seawater made  up  from hypersaline brine).

 11.1.2  Effluent  concentrations are expressed in percent effluent.

 11.1.3  Eight  replicate test vessels, each containing five 7-day old
 animals, are used  per effluent concentration and control.

 11.2   TEST  SOLUTIONS

 11,2.1  Surface waters

 11.2.1.1  Surface  water toxicity  is determined with samples  used directly as
 collected.

 11.2.2  Effluents

 11.2.2.1  The  selection of the effluent test concentrations  should be based
 on the objectives  of the study.   One of two dilution factors,  approximately
 0.3 or 0.5, is commonly used.  A  dilution factor of approximately 0.3 allows
 testing between  100% and 1% effluent using only five effluent  concentrations
                                      184

-------
000%, 30%, 10%, 3%, and UK  This series of dilutions minimizes the level
If effort, but because of the wide interval between test concentrations
 rovides poor test precision.  A dilution factor of 0.5 provides
Greater precision, but requires several additional dilutions^ span the
;ame range of effluent concentrations.  Improvements in precision decline
•apidly as the dilution factor is increased beyond 0.5.

 1222   If the effluent is known or suspected to be highly toxic, a lower
 •ange of effluent concentrations should be used (such as 10%, 3%, 1%, 0.3%,
 ,nd 0.1%).  If high mortality is observed during the first 1 or 2 h of the
 •est, additional concentrations can be added.

 -i 2 2 3   The volume of effluent required for daily renewal of eight
 Implicates per concentration, each containing 150 mL of test solution, is
 Spproximately 1200 mL.  Prepare enough test solution (approximately 1600 mL)
 it each effluent concentration to provide 400 mL additional volume for
 :hemical  analyses.

 1224   The test  should begin as soon as possible, preferably within 24  h
fafter sample collection.  In no case  should the test be started more than
72 h  after sample collection.  Oust, prior to testing,  the  temperature of the
fsample  should be  adjusted to the test temperature  (26  - 27°C) and
 laintained at that  temperature while  the dilutions  are being made.

Ill  2.2.5   Effluent  dilutions should be prepared for all replicates  in each
treatment in one  flask  to ensure  low  variability  among the replicates.  The
Kelt  chambers  (cups)  are  labelled  with the test concentration and  replicate
[number    Dispense 150 mL of the appropriate  effluent dilution to each cup.

!n.3  START  OF  THE  TEST

[11.3.1   Begin  the test  by randomly placing five  animals  (one at a  time)  in
 each test cup  of  each treatment  using a  large  bore (4  mm  ID)  pipette.  It  is
!eas er fo ca ture the animals  if  the  volume  of  water  in the dish  is reduced
[and the dish is placed  on  a light table.   It is  recommended that the
!??ansflr pipette  be rinsed  frequently because  mysids  tend to adhere to  the
 inside surface.

 11.4  LIGHT, PHOTOPERIOD,  DO,  AND TEMPERATURE

HI.4.1  The light quality and intensity under ambient laboratory conditions
 are generally adequate.  Light intensity of 10-20 uE/mZ/s, or 50 to  00
 foot candles (ft c), with a 16 h light and 8 h dark cycle and a 30 mm
fphase- n  out per od is recommended.    It is critical that the test water
 temperature be maintained at 26 - 27<>C.  It is recommended that the test
fwater temperature be continuously recorded.

 11 4 1 1  If a water bath  is used to  maintain the test temperature, the
 water depth surrounding the test cups should be at least  2.5 cm deep.
                                       185

-------
  11.4.1.2  Rooms or incubators with high volume ventilation should be used
  with caution because the volatilization of the test solutions and
  evaporation of dilution water may cause wide fluctuations in salinity.
  Covering the test cups with clear polyethylene plastic may help prevent
  volatilization and evaporation of the test solutions.

  11.4.2  Low DOs may be a problem when conducting effluent toxicity tests.
  However, test cups should not be aerated unless  the DO falls below 60% of
  saturation.   The higher concentrations of some effluents  will  require
  aeration to maintain  adequate DO concentrations.  If one solution  is  aerated
  then all the treatments and the  controls must  also  be  gently aerated.

  11.5  FEEDING

  11.5.1   During  the test,  the  mysids  in  each  test chamber  should be fed
  Arteima  nauplii, which  are  less  than  24-h  old, at the  rate of  150 nauplii
  per  mysid per day.  Adding  the entire daily  ration  at  a single feeding
  immediately  after  test  solution  renewal  may  result  in  a significant DO
  depression.  Therefore,  it  is preferable to  feed half  of the daily ration
  immediately  after  test  solution  renewal, and the second half 8 - 12 h
  later.   Increase the feeding  if  the nauplii are consumed in  less than 4 h.
  It is  important that the nauplii be washed before introduction to the test
  vessels.

  11.6  TEST SOLUTION RENEWAL

  11.6.1  Test solutions are renewed daily.  Slowly pour  off all but 10 cm of
 the old test medium into a 20 cm diameter culture dish  on  a light table.  Be
 sure to check for animals that may have adhered to the  sides  of the test
,vessel.  Rinse them back into the test cups.   Add 150 mL of new test
 solution slowly to each cup.  Check the culture dish for animals that may
 have been poured out with the old media, and  return  them to the test
 vessel.

 11.7  ROUTINE CHEMICAL  AND PHYSICAL DETERMINATIONS

 11.7.1   At a  minimum,  the following measurements  should be  made in  at  least
 one replicate in the control and  the  high and low test  concentrations  at the
 beginning of  the test:  temperature,  dissolved oxygen, pH,  and salinity (see
 Figure  14).

 11.7.2   DO should  be measured  in  at  least one replicate in  the  control  and
 the high  and  low test concentrations before renewing the test medium and
 after the medium is renewed  each  day.   In addition to the daily
 calibrations, the DO meter should be checked  at least once  a week against a
 standard  Winkler titration.                       -    ;

 11,7.3  pH, temperature,  and salinity should  be measured in at  least one
replicate for each  treatment at the beginning of each 24-h exposure period.

 11.7.4  It may be advisable  to measure the  ammonia and  nitrite in the
controls  before  each renewal to be certain  that toxicity from these sources
is  not confounding the test results.
                                      186

-------
  11.8   OBSERVATIONS  DURING  THE  TEST

  11.8.1 The  number of  live  mysids are counted and recorded each day when the
  test solutions are  renewed  (see Figure  15).  Dead animals and excess food
  should be removed with a pipette before the test solutions are renewed.

  11.9  TERMINATION OF THE TEST

  11.9.1  After measuring the DO, pH, temperature, and salinity and recording
  survival, terminate the test by pouring off the test solution in all the
  cups to a one-cm depth and refilling the cups with clean seawater.  This
 will keep the animals alive, but not exposed to the toxicant, while waiting
 to be examined for sex and the presence of eggs.

  11.9.2  The live animals must be examined for eggs and the sexes determined
 within 12 h of the termination of the test.  If the test was  conducted in
 the field,  and the animals cannot be examined on site,  the live  animals
 should be shipped back to the laboratory for processing.  Pour each
 replicate into a labelled 100 ml plastic screw-capped jar,  and send  to the
 laboratory  immediately.

 11.9.3  If  the test  was  conducted  in the laboratory,  or  when  the  test
 animals arrive in the  laboratory from the  field test  site,  the test
 organisms must be processed immediately  while  still alive  as  follows:

 11.9.3.1  Examine each replicate under a stereomicroscope  (240X)  to
 determine the  number of  immature animals,  the sex of  the mature animals, and
 the presence or absence of  eggs  in the oviducts or brood sacs  of  the females
 (see Figures 2-5). This must  be  done while  the  mysids are alive because they
 turn opaque  upon  dying.   This  step should  not be attempted by  a person  who
 has not had  specialized training in  the  determination of sex and  presence of
 eggs in the  oviduct.

 11.9.3.2  Record  the number of  immatures, males, females with eggs and
 females without eggs on data sheets.

 11.9.3.3  Rinse the mysids by pipetting them into a small netted  cup and
 dipping the  cup into a dish containing deionized water.  Using forceps,
 place the mysids  from each replicate cup on tared weighing boats   and dry at
 60°C for 24  h or  at  105<>C for at least 6 h.

 11.9.3.3.1   Pieces of aluminum foil  (1-cm square) or small  aluminum weighing
 boats can be used for dry weight analyses.   The weighing pans  or  boats
 should not exceed 10 mg in weight.

 11.9.3.3.2   Number each pan with a waterproof pen with the  treatment
concentration and replicate number.   Forty-eight (48)  weighing pans are
required per test if all  the organisms survive.
                                     187

-------
               MATURE FEMALE,EGGS  IN OVIDUCTS
                      eyestalk
antennule
          antenna
                               carapace
 [\* V* Ttm^ -^wstotocyst
* developing ^x%Sg^  .  .
     broad    \T  ^^^-telson
     sac    P^opods
                                    'developing  brood sac

                                  oviducts with  developing  ova
            Figure 2.  Mature female M. bah la with eggs in oviducts,
                      From Lussier, Kuhn, and Sewall, 1987.
                                   188

-------
              MATURE  FEMALE,  EGGS  IN BROOD  SAC
                      eyestolk
:antennule
           ontenno
                                coropace
                        brood sac with
                         developing embryos
siotocyst

    telson
                                                  uropod'
                                     brood soc with
                                       developing embryos
                                   oviducts with developing ovo
   Figure 3. Mature female M.  bahia with eggs in oviducts and developing
            embryos  in the brood  sac.  Above:  lateral view.  Below: dorsal
            view.  From Lussier,  Kuhn, and Sewall, 1987.
                                  189

-------
                           MATURE  MALE
                      eyestotk
                               carapace
antennule
                                                        statocyst

                                                            lelson
                                   gonad
                                  oil globules
         Figure  4.  Mature male M. bahia.  From  Lussier, Kuhn,
                   and Sewall, 1987.
                                  190

-------
                             IMMATURE
ontennule
          antenna
                      eyestolk
                               corapoce
slotocyst

   telson
                                                  uropod'
   Figure  5. Immature M.  bahia, (A) lateral view, (B)  dorsal view.
            From LussTerT^ufrn, and Sewall, 1987.
                                   191

-------
 11..9.3.3.3  Remove the pans from the oven and transfer immediately to a
 dessicator.  After cooling for 1 h, weigh to the nearest microgram.

 12.  ACCEPTABILITY OF TEST RESULTS

 12.1  The minimum requirements for an acceptable test are 80% survival and
 an average weight of at least 0.20 mg/mysid in the controls.   If fecundity
 in the controls is adequate (egg production by 50% of females),  fecundity
 should be used as a criterion of effect in addition to survival  and growth.

 13.  SUMMARY OF TEST CONDITIONS

 13.1  A summary of test conditions is listed in Table 2.

 14.  DATA ANALYSIS
 14.1   GENERAL

 14.1.1   Tabulate and summarize  the  data.
 survival,  growth,  and fecundity data.
Table 3 presents a sample set of
 14.T.2  The  end  points  of  the  mysid  7-day rapid-chronic  test  are  based  on
 the  adverse  effects  on  survival,  growth,  and  egg  development.   Point
 estimates, such  as LCI,  LC5, LC10 and  LC50, are calculated  using  Probit
 Analysis (Finney,  1971).   LOEC and NOEC values, for  survival,  growth, and
 reproduction are obtained  using a hypothesis  test  approach  such as  Dunnett's
 Procedure (Dunnett,  1955)  or Steel's Many-one Rank Test  (Steel, 1959;
 Miller,  1981).   See  the  Appendix  for examples of  the manual computations,
 program  listings, and examples of data input  and  program output.

 14.1.3  The  statistical  tests  described here  must  be used with  a  knowledge
 of the assumptions upon  which  the tests are contingent.  The  assistance of  a
 statistician is  recommended for analysts  who  are  not proficient in
 statistics.

 14.2 EXAMPLE OF  ANALYSIS OF MYSID SURVIVAL DATA

 14,2.1   Formal statistical analysis  of the survival  data is outlined  in
 Fiqure 6.  The response  used in the  analysis  is the  proportion of animals
 surviving  in each test or  control  chamber.  Separate analyses are performed
,for  the  estimation of the  NOEC and LOEC end points and for  the estimation of
 the  LCI,  LC5, LC10 and LC50 end points. Concentrations at which there is no
 survival  in  any  of the test chambers are  excluded  from statistical  analysis
 of the NOEC  and  LOEC, but  included in  the estimation of  the LC end  points.

 14.2.2  For  the  case of  equal  numbers  of  replicates  across  all
 concentrations and the control, the  evaluation of  the NOEC  and LOEC end
 points is made via a parametric test,  Dunnett's Procedure,  or a
 nonparametric test, Steel's Many-one Rank Test, on the arcsin transformed
 data.  Underlying assumptions  of  Dunnett's Procedure, normality and
 homogeneity  of variance, are formally  tested.  The test  for normality is the
 Shapiro-Wilks Test, and  Bartlett's Test is used to determine the homogeneity

                                      192

-------
   TABLE 2. SUMMARY OF RECOMMENDED TEST CONDITIONS FOR MVSIDOPSIS BAH IA
            SEVEN DAY SURVIVAL, GROWTH, AND FECUNDITY TEST
 1.  Test  type:
 2.  Salinity:
 3.  Temperature:
 4.  Photoperiod:

 5.  Light intensity:
 6.  Test  chamber:

 7.  Test  solution  volume:
 8.  Renewal  of  test  solutions:
 9.  Age of test organisms:
10.  Number of treatments  per  study:

11.  Number of organisms per  test
     chamber:
12.  Number of replicate chambers
     per treatment:
13.  Source of food:
14.  Feeding regime:

15.  Aeration:
16.  Dilution water:
17.  Test duration:
18.  Dilution factor:
19.  Effects measured:
20.  Cleaning:
Static renewal
20 o/oo to 30 °/oo + 2 "Voo
26 - 27°C
16 h light, 8 h dark, with phase
in/out period
10-20 uE/m2/s (50-100 ft,c,)
8 oz plastic disposable cups, or
400 mL glass beakers
150 ml per replicate cup
Daily                , ;^;
7 days
Minimum of 5 treatments a-nd a
control
8
Artemia nauplii
Feed 150 24-h old nauplii per mysid
daily, half after test solution
renewal and half after 8 - 12 h.
None unless DO falls below 60%
saturation, then gently in all cups
Natural sea water or hypersaline brine
7 days
Approximately 0.3 or 0.5
Survival, growth, and egg development
Pipette excess food from cups daily
                                    193

-------
  TABLE 3. DATA FOR MYSID SHRIMP 7-DAY SURVIVAL, GROWTH, AND FECUNDITY TEST
Treatment Replicate
Chamber
-- — ~ — — 	 —
1
i
2
£_
3
v
Control 4
5
•j
6
\f
7
1
8
1
i
2
£»
3
*j
50 ppb 4
6
w
7
/
8
1
i
2
L»
3
^
100 ppb 4
g
U
7
/
8
1
i
~
210 ppb 4
A
u
7
/
8
1
i
2
3

-------
of variance.  If either of these tests fail, the nonparametric test, Steel's
Many-one Rank Test, is used to determine the NOEC and LOEC end points.  If
the assumptions of Dunnett's Procedure are met, the end points are estimated
by the parametric procedure.

14.2.3   If unequal numbers of replicates occur among the concentration
levels tested, there are parametric and nonparametric alternative analyses.
The parametric analysis is the Bonferroni t-test.  The Wilcoxon Rank Sum Test
with the Bonferroni adjustment is the nonparametric alternative.  For
detailed information on the Bonferroni adjustment see the Appendix.

14.2.4  Probit Analysis (Finney, 1971) is used to estimate the concentration
that causes a specified percent decrease in survival from the control.  In
this analysis, the total mortality data from all test replicates at a given
concentration are combined {total number dead at concentration level i
divided by total number exposed at concentration level i),

14.2.5  The proportion of survival in each replicate must first be
transformed by the arcsin transformation procedure decribed in the Appendix.
The raw and transformed data,  means and standard deviation of the transformed
observations at each concentration including the control  are listed in
Table 4,  A plot of the mean survival is provided in Figure 7.

                    TABLE 4.  MYSIDOPSIS  BAHIA  SURVIVAL DATA
                                         Concentration
       Replicate   Control
50.0
100.0
210.0
450.0



Raw





Arcsin
Trans
formed




Mean(Yi)
Si2
i
i
1
2
3
4
5
6
7
8
1
2
3
4
5
6
7
8



0.80
0.80
1.00
1.00
1.00
1.00
1.00
0,80
1.107
1.107
1.345
1,345
1.345
1.345
1.345
1.107
1.256
0.015
1
0.80
1.00
0.80
0.80
1.00
1.00
0.80
1.00
1.107
1.345
1.107
1,107
1.345
1.345
1.107
1.345
1.226
0.016
2
0.60
1.00
1.00
1.00
1.00
0.60
0.80
0.80
0.886
1.345
1.345
1.345
1,345
0.886
1.107
1.107
1.171
0.042
.3
1.00
0.80
0.20
0.80
0.60
0.80
0.80
0.80
1.345
. 1,107
0.464
1.107
0,886
1.107
1.107
1.107
1.029
0.067
4
0.00
0.20
0.00
0.20
0.00
0.00
0.00
0.40
0.225
0.464
0.225
0.464
0.225
0.225
0.225
0.685
0.342
0.031
5
                                      195

-------
1
I
•
STATIST
SURV]
*
PROBIT
ANALYSIS
1
ENDPOINT ESTIMATE
LCI, LC5, LC10.LC50
NORMAL
HOMOGENEOUS VARIANCE
™ 	 EQUAL
REF
ICAL ANALYSIS OF MYSIDOPSIS BAHIA
•VAL. GROWTH AND FECUNDITY TEST
SURVIVAL
SURVIVAL DATA
PROPORTION SURVIVING
1
f
ARCSIN
TRANSFORMATION

	 I 	 NUN-NUHM

DISTRIBUTION!
' i Hfc
BARTLETT S TEST » •**
1 i
i t
. NUMBER OF EQUAL NUMBER OF
1ICATES? REPLICATES?
YES I ' I YES
T-TEST WITH n...
BONFERRONI UUP
ADJUSTMENT




1 \ \ WTI rr
JNETT'S1 STEEL'S MANY-ONE ' "^^
TEST j RANK TEST | BONFERT

AL DISTRIBUTION
TEROGENEOUS
VARIANCE
NO
v
XON RANK SUM
EST WITH
ONI ADJUSTMENT

•'"-• i
ENDPOINT ESTIMATES
NOEC, LOEC t

Figure 6. Flow chart for analysis of mysid survival  data



          ••?•>•             196

-------
z
o
o
o
o
a:
o
U4
UJ
X
o
o
o
C
O)
E
-M
(O
0)
                                                                                                  O
                                                                                                  fTJ
                                                                                                  O)
                                                                                                  -o
                                                                                                  1.
                                                                                                  3
                                                                                                  V)
                                                                                                  o

                                                                                                 0.
                                                                                                  cu
                                                                                                  i-
                                       NOIiaOdOMd
                                              197

-------
|4.2.6   Test  for  Normality          <-^ .,

114.2.6.1  The  first  step of the test for normality is to center the
Observations  by subtracting the mean of all observations within a
 :oncentration from  each observation in that concentration . The centered
^observations  are  listed in Table 5.
            TABLE 5.   CENTERED OBSERVATIONS FOR  SHAPIRO-WILKS  EXAMPLE
El' 1 flLJ L~ L* sJ* ^t^JllL«iM«LJ \JLJOL.l\Vr1FlWH*J 1 UP\ *J J L/il 1 IV W H .1 L* T\ *J L- r\r\\ n L. L.
1
1 Concentration
1
I Replicate

1
2
3
4
5
6
7
8
Control
{Site Water)
-0.149
-0.149
0.089
0.089
0.089
0.089
0.089
-0.149
50.0

-0.119
0.119
-0.119
-0.119
0.119
0.119
-0.119
0.119
100.0

-0.285
0.174
0.174
0.174
0.174
-0.285
-0.064
-0.064
210.0

0.316
0.078
-0.565
0.078
-0.142
0.078
0.078
0.078
450.0

-0.117
0.121
-0.117
0.121
-0.117
-0.117
-0.117
0.342
14.2.6.2  Calculate the denominator, D, of the test statistic:

                         n
                     D = t (Xi - X)2
    Where Xj = the ith centered observation
          x  = tne overall mean of the centered observations
          n  = the total number of centered observations.
For this set of data,
n = 40
                                      1  (-0.006)  =  0.0
                                     W
                                 D = 1.197       ?   -

14.2.6.3  Order the centered observations from smallest  to  largest:

                  x(l)  . x<2) -  ...  - x(n)

Where %d} is the ith ordered observation.  These ordered observations
are listed in Table 6.
                                    198

-------
     TABLE 6.  ORDERED CENTERED OBSERVATIONS FOR SHAPIRO-WILKS EXAMPLE
                                                         x(D
1
2
3
4
• ?,. 5
•;;i 6
7
8
9
10
n
12
13
14
15
16
17
18
19
20
-0.565
-0.285
-0.285
-0.149
-0.149
-0.149
-0.143
-0.119
-0.119
-0.119
-0.119
-0.117
-0.117
-0.117
-0.117
-0.117
-0.064
-0.064
0.078
0.078
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
0.078
0.078
0.078
0.089
0.089
0.089
0.089
0.089
0.119
0.119
0.119
0.119
0.121
0.121
0.174
0.174
0.174
0.174
0.316
0.342
14.2.6.4  From Table 4, Appendix B,  for the number  of observations,  n,
obtain the coefficients a],  a2»..--> aR where k  is  approximately
n/2. For the data in this example,  n = 40 and k  = 20.  The  a-,*  values
are listed in Table 7.

14.2.6.5  Compute the test statistic,  W,  as follows:
                   1   k
                   D  1 = l'
                                      .  X(D)  ]2
The differences X
-------
 14.2.6.6 The decision rule for  this  test  is  to  compare  W  with  the
 critical  value  found in  Table 6,  Appendix B.   If  the  computed  W  is  less
 than the critical  value,  conclude that  the data are not normally
 distributed.

 For  this set of data,  the critical value  at  a significance  level of  0.01
 and  n =  40  observations  is 0.919.  Since  W = 0.9167 is  less  than the
 critical  value,  the  conclusion  of the test is that it is  reasonable  to
 assume the  data  are  not  normally  distributed.

 14.2.6.7 Since  the data  do not  meet  the assumption of normality, Steel's
[Many-One  Rank Test will  be used to analyze the  survival data.
      TABLE 7.  COEFFICIENTS AND DIFFERENCES FOR SHAPIRO-WILKS EXAMPLE
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
0.3964
0.2737
0.2368
0.2098
0.1878
0.1691
0.1526
0.1376
0.1237
0.1108
0.0986
0.0870
0.0759
0.0651
0.0546
0.0444
0.0343
0.0244
0.0146
0.0049
0.907
0.601
0.459
0.323
0.323
0.323
0.264
0,240
0.238
0.238
0.238
0.236
0.206
0.206
0.206
0.206
0.153
0.142
0.0
0.0
X(40)
X(39)
X<38)
X(37)
X 36)
X 35
X(34)
X(33)
x(32)
x(31 }
X<30)
X(29)
x(28)
X(27)
X<26)
X(Z5)
X(24)
X<23)
X(22)
X(21)
- x(D
- x 2>
- X(3)
- * 4!
- X<5)
- x(6)
- x(7)
- X<8)
- X<9)
- xdo)
- xdi)
- X(12)
- X(13)
- X(14)
- X(15)
- X<16)
- xH7)
- X(18)
- X(19)
- X(20)
                                    200

-------
14.2.7  Steel's Many-One Rank Test

14.2.7.1  For each control and concentration combination, combine the
data and arrange the observations in order of size from smallest to t^.
largest.  Assign the ranks (1,2, ... ,40) to the ordered observations"
with a rank of 1 assigned to the smallest observation, rank of 2 assigned
to the next larger observation, etc.  If ties occur when ranking, assign
the average rank to each tied observation.

14.2.7.2 An example pf assigning ranks to the combined data for the
control and 50.0 ppb concentration is given in Table 8. This ranking
procedure is repeated for each control/concentration combination.  The
complete set of rankings is summarized in Table 9.  The ranks are next
summed for each concentration level, as shown in Table 10.

1.4.2.7.3 For this example, we want to determine if the survival in any
of the concentrations is significantly lower than the survival in the
control. If this occurs, the rank sum at that concentration would be
significantly lower than the rank sum of the control.  Thus we are only
concerned with comparing the rank sums for the survival at  each of the
various  concentration levels with some "minimum" or critical rank sum,
at or below which the survival would be considered significantly lower
than the control.  At a significance level of 0.05, the minimum rank sum
in a test with four concentrations (excluding the control)  and eight
replicates is 47 (See Table 5, Appendix E).

14.2.7.4  Since the rank sum for the 450 ppb concentration  level is less
than the critical value, the proportion surviving in that concentration
is considered significantly less than that in the control.   Since no
other rank sums are less than or equal to the critical value, no other
concentrations have a significantly lower proportion surviving than the  ,
control.  Hence, the NOEC and the LOEC are assumed to be 210.0 ppb and   /
450.0 ppb, respectively.                                                t
                                    201

-------
   TABLE 8.  ASSIGNING RANKS TO THE CONTROL AND 502 CONCENTRATION LEVEL
                      FOR STEEL'S MANY-ONE RANK TEST
         Rank
Transformed Proportion
  of Total Mortality
Concentration
4
4
4
4
4
4
4
12 :
12
12
12
12
12
12
12
12
1.107
1.107
1.107
1.107
1.107
1.107
1.107
1.571
1.571
1.571
1.571
1.571
1.571
1.571
1.571
1.571
Control
Control
Control
50%
50%
50%
50%
Control
Control
Control
Control
Control
50%
50%
50%
50%
                         TABLE  9.   TABLE  OF  RANKS^
MRepl
i- . Control
50
100
210
450
n cate
HuR

1 ?
11 i
i
IB 6

1 8
1.107(4,5,6
1.107(4,5,6
1.345(12,12
1.345(12,12
1.345(12,12
1.345(12,12
1.345(12,12
1.107(4,5,6
.5,10)
.5,10)
,13.5,14)
,13.5,14)
,13.5,14)
,13.5,14)
,13.5,14)
.5,10)
1.107(4)
1.345(12)
1.107(4)
1.107(4)
1.345(12)
1.345(12)
1.107(4)
1.345(12)
0.886(1.5)
1.345(12)
1.345(12)
1,345(12)
1.345(12)
0,886(1.5)
1.107(5)
1.107(5)
1.345(13.5)
1.107(6.5)
0.464(1)
1,107(6.5)
0.886(2)
1.107(6,5)
1.107(6.5)
1.107(6.5)
0.225(3)
0.464(6.5)
0.225(3)
0.464(6.5)
0.225(3)
0.225(3)
0.225(3)
0.685(8)
^Control  ranks are given in the order  of  the  concentration with which
 they were ranked.
                                    202

-------
                            TABLE 10.  RANK SUMS
Concentration
50
100
210
450
Rank Sum
64
61
49
36
14.2.8  Probit Analysis
14.2.8.1
Table 11.
Program.
Figure 8.
The data used for the probit analysis is summarized in
 To perform the probit analysis, run the EPA Probit Analysis
An example of the program output is provided in Table 12 and
14.2,8.2 For this example, the chi-square test for heterogeneity was not
significant.  Thus probit analysis appears to be appropriate for this set
of data.
                    TABLE 11.   DATA FOR PROBIT ANALYSIS

No. Dead
No. Exposed
Control
3
40

50.0
4
40

Concentration
100.0 210.0
6
40
11
40

450.0
36
40
                                   203

-------
 TABLE  12.  OUTPUT FROM  EPA PROBIT  ANALYSIS  PROGRAM,  VERSION  1.3,
            USED FOR CALCULATING  EC VALUES
Prpbit Analysis of Hysidopsis Bah


Con c ,
y Control
50.0000
f 100.0000
' 210 . 0000
450 . 0000
Chi ~ Square
Mu
Sigma
Parameter
Intercept
Elope '

Number Number
Exposed Resp.
40 4
40 4
40 6
40 1 1
40 , 36
Heterogeneity = 0
2 . 4641 1 7
0 . 156802
Estimate Std.
-10.714873 3.194
6.377486 1.274
ia Survival
Observed
Proportion
Respond ing
0 . 1000
0 . 1000
0 . 1500
0 2750
0. 9000
. 51 7


Err
243 ( - 1
491 (
Data
Adjusted
Pr opor t i on
Responding
0. 0000
- . 0179
0 . 0387
0.1801
0. 8869

•

95 % Con f i den ce
6 91 5550 , -4
3879483. 8

Predicted
Proportion
Re s pond ing
0.1158
0 . 0000
0 . 001 5
0.1827
0 . 8861



Limits
454156)
875488)
Spontaneous     0.115787

Response Rate
0.029555
0.057859,
0.173715)
      Estimated EC Values and Confidence Limits
Point
                   Cone
                                      L ower        U~ppe r

                                    95% Confidence Limits
EC 1
EC 5 ,
EC1 0 .
EC15 .
EC50.
EC85.
EC90 .
EC95 .
EC99 .
00
. 00
00
00
00
00
00
00
00
">;'' 125
160
183
200 .
291 .
423
462 .
527 .
674 .
7042
7661
2992
2664
1 503
2791
4602
2789
3496
66 .
98
1 21
139
240.
361
391 .
436 .
528 .
2441
3363
0638
0534
4525
8588
0961
0173
6572
169
203
225
242
338
545.
621 .
760 .
1121.
1 21 1
7259
' 5930
0966
8918
0946
6678
1 824
1 471
                                    204

-------
           PLOT OP  ADJUSTED PROBITE AND PREDICTED REGRESSION LINE
     Probit
       10*
        7*
         EC 01
                     EC10     EC25     EC50     EC75    EC90
                                                                EC99
Figure 8.  Plot  of adjusted probits and  predicted regression  line.
                                 205

-------
 14.3   EXAMPLE  OF ANALYSIS  OF MYSID GROWTH DATA

 14.3.1  Formal  statistical  analysis of  the growth data  Is outlined  in
 Figure  9.  The  response used in the statistical analysis is mean weight of
 males  and females combined  per replicate.  Concentrations above the NOEC for
 survival are excluded from  the growth analysis.

 14.3.2  The statistical analysis consists of a parametric test, Dunnett's
 Procedure, and  a non-parametric test, Steel's Many-one Rank Test. The
 underlying assumptions of the Dunnett's Procedure, normality and homogeneity
 of variance, are formally tested.  The test for normality is the
 Shapiro-Wilks Test and Bartlett's Test is used to test for homogeneity of
 variance.  If either of these tests fail, the non-parametric test, Steel's
 Many-one Rank Test, is used to determine the NOEC and LOEC end points.  If
 the assumptions of Dunnett's Procedure are met, the end points are
 determined by the parametric test,       - '

 14.3.3  Additionally,  if unequal  numbers of replicates occur among the
 concentration levels tested there are parametric and non-parametric
 alternative analyses.   The parametric analysis is the Bonferroni t-test.  The
 Wilcoxon Rank Sum Test with the Bonferroni adjustment is the non-parametric
 alternative.   For detailed information on the Bonferroni adjustment, see  the
Appendix.

 14.3,4  The data, mean and standard deviation of the observations at each
concentration including  the control  for  this  example are listed in
Table 13.   A plot of the data  is  provided in  Figure  10.   Since  there is
significant mortality  in the 450  ppb  concentration,  its  effect  on growth  is
not considered,

                         TABLE  13,  MYSID GROWTH DATA
Replicate   Control
                                           Concentration  (ppb)
                                 50.0
100.0
210.0
450.0








Mean(Yi)
Si2
i
1
2
3
4
5
6
7
8



0.183
0.148
0.216
0.199
0.176
0.243
0.213
0.180
0.195
0.0008
1
0.192
0.193
0,237
0.237
0.256
0,191
0.152
0.177
0.204
0.0012
2
0.190
0.172
0.160
0.199
0.165
0.241
0.259
0.186
0.197
0.0013
3
0.153
0.117
0.085
0.153
0.086
.'• 0.193
•='• 0.137
0.129
0.132
0.0013
4
_
0.060
-
0.009
-
_
-
0.203

-
5
                                     206

-------
STATISTICAL ANALYSIS OF MYSIDOPSIS BAHIA
SURVIVAL. GROWTH AND FECUNDITY TEST
GROWTH
, • GROWTH DATA
MEAN WEIGHT
(EXCLUDING CONCENTRATIONS ABOVE NOEC FOR SURV


bnArlHU nILIso ILof
NORMAL DISTRIBUTION 1
. ^ HE1
BARTLETT S TEST 	 • 	 -^*
HOMOGENEOUS VARIANCE
£5 	 EQUAL NUMBER OF EQUAL NUMBER OF
REPLICATES? REPLICATES?
VPQ YF9
V 	 , j |
T-TEST WITH ni|NKIFTT.s STEEL'S MANY-ONE WILCD>
BONFERRONI DUNTN|STTT S RANK TEST TE
ADJUSTMENT TfcS' 	 HA . 	 	 BONFERRC

" 1
ENDPOINT ESTIMATES
NOEC, LOEC
IVAL}
L DISTRIBUTION
FEROGENEOUS
VARIANCE
NO
v
ON RANK SUM
.ST WITH
)NI ADJUSTMENT




Figure 9.  Flow chart for statistical  analysis  of  mysid  growth  data
                                  207

-------
  (/I
ZUJ
Ol-

t-tn
       O
       O:
LUZ-JI—
Z3OO
OOSO
O
CJU.— lt-
<  <_
UJLU>3
  3   O
U_>X
    >
UJ—J   Z

-JOOQ;
  — 00 U.
ZQS  —
LU  X
3bJO>-
  x —_*
LUI—LUh-
      :z
      •<
I— LULU U.
OV13 —
LULU
ZO.Z —
OLU<(/»
                                                                                                        O
                                                                                                        s~
                                                                                                        Ol
                                                                                                        O
                                                                                                       **-
                                                                                                        ui
                                                                                                        
                                                                                                        •r-
                                                                                                        U.
                                                      208

-------
  14.3.5  Test for Normality
 JI4.3.5.1  The first step of the test for normality  is to center the  J
 Sobservations by subtracting the mean of all observations within a
 fconcentration from each observation in that concentration.  The centered
 Sobservations are listed in Table 14.
           TABLE 14.  CENTERED OBSERVATIONS FOR SHAPIRO-WILKS EXAMPLE
||H Concentration
mm
K Replicate
^fflp
• 1 •
8B» %
H 3
• 4
• 5
H 6
H 7
H 8
RHP . .... .,_
Control
-0.012
-0.047
0.021
0,004
-0.019
0.048
0.018
-0.015
50.0
-0.012
-0.011
0,033
0.033
0.052
-0.013
-0.052
-0.027
100.0
-0.006
-0,024
-0.036
0.002
-0.032
0.044
0.062
-0.010
(ppb)
210.0
0.021
-0.015
-0.047
0.021
-0.046
0.061
0.005
-0.003
;14.3.5.2  Calculate the denominator, D, of the statistic:
                           n       _
                       D = I (Xi  - X)2
                          i=l
= the ith centered observation
= tne overall mean of the centered observations
     Where
             n  = the total  number of centered observations
[14.3.5.3    For this set of data:  n =  32
                                    X =  -   (-0.000)  =  0.000
                                    D =  0.0329
 14.3.5.4     Order  the centered  observations from smallest  to largest
                X(l)  - X(2)  -  ...  -  X(n)
 /here  x(i)  denotes the ith  ordered  observation.   The  ordered
Pobservations for this example are listed  in Table 15.
                                     209

-------
   TABLE  15
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
15
16
            ORDERED CENTERED  OBSERVATIONS FOR  SHAPIRO-WIUS  EXAMPLE
            -0.052
            -0.047
            -0.047
            -0.046
            -0.036
            -0.032
            -0.027
            -0.024
            -0.019
            -0.015
            -0.015
            -0.013
            -0.012
           -0.012
           -0.011
           -0.010
                                         17
                                         18
                                         19
                                        20
                                        21
                                        22
                                        23
                                        24
                                        25
                                        26
                                        27
                                       28
                                       29
                                       30
                                       31
                                       32
 -0.006
 -0.003
 0,002
 0.004
 0.005
 0.018
 0,021
 0.021
 0.021
 0.033
 0.033
 0.044
 0.048
0.052
0.061
0.062
|14.3.5.6  Compute the test statistic,  W,  as follows:

                    1    k
                "  ~ —  L  Z  a -i  (x(n~7+l)   vM ) \  to
                   D   1 — 1  '             A*'')JC.
\
The differences x(n-i + l) . y(i) avio ,. .   .  .
in this examolP.           X    are llsted  m Table 16.
example,
                                                   For the data

           0.0329
                              = 0.9469
                             210

-------
    TABLE 16.  COEFFICIENTS AND DIFFERENCES FOR SHAPIRO-WILKS EXAMPLE
J .— . — M.^ — . — .^ .^.- .«-_ | ,|
1 0.4188
[ 2 0.2898
1 3 0.2462
4 0.2141
1 5 0.1878
1 6 0.1651
1 7 0.1449
8 0.1265
1 9 0.1093
: . 10 0.0931
11 0.0777
12 0.0629
13 0.0485
14 0.0344
15 0.0206
16 0.0068
0.114
0.108
0.099
0.094
0.080
0.065
0.060
0.045
0.040
0.036
0.033
0.018
0.016
0.014
0.008
0.004
X(32)
X(3D
X(30)
X<29)
X(28)
X(27)
X(26)
X(25)
X(24)
X(23)
X(22)
x(21 )
X(20)
X<19)
x(18)
Xd7)
- xd)
- X<2)
- X(3)

- X(5)
- X<6)
- X(7)
- X(8)

- xdo)

- x(12)
- Xd3)

- xHs)
- X(16)
 14.3.5.7  The  decision  rule  for  this  test  is  to  compare  W  as  calculated
 in  14.3.5,6  to a  critical  value  found in Table 6,  Appendix B.   If  the
 computed W is  less  than the  critical  value, conclude  that  the  data are
 not  normally distributed.  For  this  set  of  data,  the critical  value at a
 signficance  level of  0.01  and  n  = 32  observations  is  0.904. Since  W =
 0.9496  is greater than  the critical value,  it is reasonable to assume
 that the data  are normally distributed.

 14.3.6   Test for Homogeneity of  Variance

 14.3.6.1  The test used  to  examine whether  the variation  in mean weight of
 the  mysids is  the same  across  all concentration  levels including the
.control,  is  Bartlett's  Test  {Snedecor and  Cochran, 1980).  The test
'statistic  is as follows:                 ;-

B =

[


P
{ I Vj
1=1

} In $2

C
P
-2V-
1=1

j In S-j2 ]


    Where V-j =   degrees of freedom for each copper concen-
                 tration and control, V-j = (nf - 1)

          p  =   number of concentration levels including
                 the control

                                    211

-------
            C   » 1  + (  3(p-l)H  [  z  i/Vl  -  {
      >:    In  =  loge

            1   ~  1,  2,  ..., p where p  is the  number of concentrations
                             including the control
            ni  =  the number of replicates for concentration 1.

  14.3.6.2   For the data  in this example (See Table 13), all concentrations
  including  the control have the same number of replicates (ni = 8 for
  all  i).  Thus,  Vi = 7 for all i.                           1

  14,3.6.3   Bartlett's statistic is therefore:


        B =  [(28)ln(0.0012)  - 71 ln{Sj)2]/l.06


          =  [28(-6.7254) - 7'(-27.1471)]/] .06

          =  [-188.3112 - (-190.0297)3/1,06

          =  1.621

 14.3.6.4  B is approximately distributed  as  chi-square  with  p -  1  degrees
 of freedom, when the  variances are  in fact the same.  Therefore,  the
 appropriate critical  value for this  test,  at a significance  level  of  0.01
 with three degrees  of  freedom, is 9.210.   Since  B =  1.621  is  less  than
 the critical  value  of  9.210,  conclude that the variances are  not
 different.

 14.3.7  Dunnett's Procedure
i
114.3.7.1   To obtain an estimate of the pooled  variance  for the Dunnett's
(Procedure,  construct an  ANOVA table as described  in Table 17.
                                      212

-------
                       TABLE 17,  ANOVA TABLE
m
mg Source

B Between
KtSS
m
•H Within
•
H Totdl
m
B
Kfhere p
H N
df

P - 1

N - p
N - 1
= number
= total
Sum of Squares
(SS)

SSB

SSW
SST
of concentration levels
number of observations ni
Mean Square(MS)
(SS/df)
2
SB = SSB/(p-l)
2
SW = SSW/(N-p)

including the control
I + n? ... +nn
           = number of observations in concentration i
       SSB = Z T12/n1 - Q2/N
                                         Between Sum of Squares
                P   I'M
          SST = I   I YH
               1=1 j=l

          SSW = SST - SSB
                        - Q2/N
Total Sum of Squares


Within Sum of Squares
4.3.7.2

   nl =
   T2 =
   T3 =
   T4 =
        G  = the grand total of all sample observations, G = z T-j
                                                            1 = 1
        T-j = the total of the replicate measurements for
             concentration "i"
       Yfj = the jth observation for concentration "i" (represents
             the mean dry weight of the mysids for concentration
             i in test chamber j)

      For the data in this example:

     n2 = n3 = n4 = 8
     32
     YII + Yi2 + ... + Y]8 = 1.558
         + Y22 + ... + Y28 = 1-635
         + Y32 + ... + Y38 = 1.572
         + Y42 + ... + Y48 = 1.053
G  = TI + T2 + T3 + T4 = 5.818
                                213

-------
     SSB  * I  Tj2/ni  -  G2/N
          i = l

         = JJ 8.680}  -  (5.818)2  = 0.027
            8             32
    SST = E   I
         1=1  =
                      - Q2/N
         =  1.118  -  (5.818)2  = o.060
                     32                                             :

     SSW  =  SST  -  SSB  = 0.060 - 0.027 = 0.033

     SB2  =  SSB/p=1  =  0.027/4-1 = 0.009

     SW2  =  SSW/N-p  =  0.033/32-4 = 0.001


 14.3.7.3   Summarize  these calculations in the ANOVA table  (Table 18

           TABLE  18.   ANOVA TABLE  FOR  DUNNETT'S  PROCEDURE  EXAMPLE
•
• Source
!8i
H Between
H
W Within
IB
H Total
H
df
3
28
31
Sum of Squares
(SS)
0.027
0.033
0.060
Mean Square(MS)
(SS/df)
0,009
0.001

,14.3.7.4  To perform the  individual comparisons, calculate the t
Statistic for each concentration, and control combination as follows
                                  Sw/
                                              + (I/nil
Where Yj  = mean dry weight for concentration i
          = mean dry weight for the control
          = square root of within mean sqaure
          = number of replicates for control
          = number of replicates for concentration i.

                                    214
      Sw

      "1
      "1

-------
f. -a
f _ *
.3.7.5  I.ble 19 include. ».
                                                                         the
                                                                    follows:
                              (  0.195 - 0.204  )

                          [ 0.032 /(I/a) +

                   TABLE 19.  CALCULATED T-VALUES
             Concentration (ppb)
                  50.0
                 100.0
                 210.0
             ,4.3.7.6  Since the purpose of this test ^^
             reduction in mean weight, a (o e-sided) t est

                                         !
                                                              -0.562
                                                              -0.125
                                                               3,938
                                                  significant


                                  s
calculated.
                 MSD =  d SW >/ (1/ni) + U/n)
Where: d
      S
      n
= the critical value for the Dunnett's procedure
=                                Sssaa
, the number of  replicates  in the control.
 14.3.7.8  In this example:
                  MSD = 2.15 (0-032) >/"
                     = 2.15 (0.032M0.5)
                     = 0.034
                      *   , i? At reduction in mean weight from the
 14.3.7.10 This  represents  a 17.4% reduction
 control.
                                  215

-------
 14.4 EXAMPLE OF ANALYSIS OF MYSID FECUNDITY DATA

 14.4.1   Formal  statistical  analysis of the fecundity data is outlined in
[Figure  11.   The response used in  the analysis is the proportion of females
fwith eggs  in each  test or control chamber.   If no females were present in a
Ireplicate,  a response of zero should not be used.  Instead there are no data
[available  for that replicate and  the number of replicates for that level of
[concentration or the control should be reduced by one.   Separate analyses
[are performed for  the estimation  of the NOEC and LOEC end points and for the
 istimation  of the  EC1, ECS, EC10  and EC50 end points.  The data for a
 ;oncentration are  excluded  from the statistical  analysis of the NOEC and
   •C if there no eggs were  produced in all  of the replicates in which
 •emales existed.  However,  all  data are included in the estimation of the EC
 *nd points,

 [4.4,2   For  the case of equal numbers of replicates across all
 :oncentrations  and the control, the evaluation of the NOEC and LOEC end
 )oints  is  made  via a parametric test, Dunnett's  Procedure, or a
 lonparametric test,  Steel's Many-one Rank Test,  on the arcsin transformed
Hata.   Underlying  assumptions of  Dunnett's  Procedure, normality and
homogeneity  of  variance, are formally tested.  The test for normality is the
IShapiro-Wilks Test,  and Bartlett's Test is used  to determine the homogeneity
Ipf  variance.  If either of  these  tests fail, the nonparametric test, Steel's
ilany-one Rank Test,  is used to determine the NOEC and LOEC end points.  If
ithe assumptions of Dunnett's Procedure are met,  the end points are estimated
 )y  the  parametric  procedure.

114.4.3   If  unequal numbers  of replicates occur among the concentration
 levels  tested,  there are parametric and nonparametric alternative analyses.
 "he parametric  analysis is  the Bonferroni t-test.  The Wilcoxon Rank Sum
 "est with  the Bonferroni adjustment is the nonparametric alternative.  For
 letailed information on the Bonferroni adjustment see Appendix D.
 t
 14.4.4   Probit  Analysis (Finney,  1971) is used to estimate the concentration
 lausing a  specified reduction in  fecundity measured by the proportion of
 'emales without eggs.  As in survival data, the  fecundity data from all test
 'eplicates  at a given concentration are combined, total number of females
 n'thout eggs at concentration i divided by the total number of females at
 ;oncentration i, to yield the proportion for that concentration.  Since the
 variable of  interest is the proportion of females producing no eggs, the
 proportion  of females without eggs as a natural  occurrence should be allowed
 -or in  the  analysis.  The natural infertility is estimated from the
proportion  of females without eggs in the control.  With an added adjustment
Ifor spontaneous infertility rate, the Probit Analysis is carried out as for
Ithe survival data.  A discussion  of the Probit Analysis with adjustment for
fspontaneous  response in the controls is included in Section 9 (Data
{Analysis).
                                       216

-------
,14.4.5   In  this  example,  the  proportion  of female  mysids  with  eggs  in each
^replicate  is  first  transformed  by the  arcsin  transformation  procedure
(described  in  Appendix  B.   Since the  denominator  of the  proportion of females
!with  eggs varies with  the number of  females occurring  in  that  replicate,  the
'adjustment  of the arcsin  transformation  for 0% and 100% is not used for this
'data.   The  raw and  transformed  data, means and standard deviations  of the
: transformed observations  at each test  concentration including  the control
'are  listed  in Table 20.   Since  there  is  significant mortality  in the 450  ppb
^concentration, its  effect on  reproduction  is  not considered.   Additionally,
^since no eggs were  produced by  females in  any of the replicates for the 210
   b  concentration,  it  is  not  included  in this statistical analysis  and is
(considered  a  qualitative  reproductive  effect.
           TABLE  20.   MYSID  FECUNDITY  DATA:  PERCENT  FEMALES  WITH  EGGS
        Replicate   Control
Test Concentration (ppb)

50.0     100,0     210.0


1
• RAW

1
i


IARC SINE
1 TRANS-
• FORMED


i

i
1 Meant Yi)
i$i2
li
,
2
3
4
5
6
7
8
1
2
3
4
5
5
7
8



1.00
1.00
0.67
1.00
1.00
0.80
1.00
1.00
1.57
1.57
0.96
1.57
1.57
1.12
1.57
1.57
1.44
0.064
1
0.50
0.33
0.67
-
0.40
0.50
0.25
0.33
0.78
0.61
' 0.96
-
0.68
0.78
0.52
0.61
0.71
0.021
2
0.33
0.50
0.00
0.50
0.67
0.00
0.25
~
0.61
0.78
0.00
0.78
0.96
0.00
0.52

0.52
0.147
3
0.0
0.0
0,0
0.0
0.0
0.0 ;
0.0 ^
0.0
_
-
.-:;:
':. :•'•''
"'.'•
••' ""
-

-
-
4
                                       217

-------
STATISTICAL
SURVIVAL.
ANALYSIS OF MYSIDOPSIS BAHIA
GROWTH AND FECUNDITY TEST
FECUNDITY
FECUNDITY DATA
PROPORTION OF FEMALES WITH EGGS
(EXCLUDING CONCENTRATIONS ABOVE NOEC FOR SI
1

PROBIT
ANALYSIS
1
ENDPOINT ESTIMATE s
EC1 EC5 EC10 EC50

+
ARCSIN
TRANSFORMATION

	 * 	 NON-NORM

f
NORMAL DISTRIBUTION!

HOMOGENEOUS VARIANCE
V
BARTLETT'S TEST 	 *4 HE

•^ 	 EQUAL NUMBER OF EQUAL NUMBER OF
REPLICATES? REPLICATES?
^- YES 1
T-TEST WITH niJNNFTT'£
BONFERRONI UUNTNFtclT fc
An.HIRTMFNT ' co J

1 YES
3 STEEL'S MANY-ONE WILC?
RANK TEST RONFERR




*
ENDPOINT ESTIMATES
NOEC, LOEC
RVIVAL)
AL DISTRIBUTION
TEROGENEOUS
VARIANCE
NO
_ — . — -
V
XON RANK SUM
EST WITH
ONI ADJUSTMENT



Figure 11.  Flow  chart  for  statistical analysis of mysid fecundity data.
                                   218   .-:vr;;:V'              .	^^

-------
                                                           a;


                                                          JC
                                                           t/j
                                                          -a
                                                           (0


                                                           O)
                                                          c:
                                                          o
                                                          S-
                                                          o
                                                          a.
                                                          o
                                                          s-
                                                          a.
                    o

S003 HUM S3TW3J NOIlMOdOMd
o
o
             219

-------
  14.4.6  Test for Normality
I 14.4.6.1  The first step of the test for normality is to center the
  observations by subtracting the mean of all observations within a
  concentration from each observation in that concentration.   The centered
  observations are listed in Table 21.
1| lAbLh 21. UtNlLKLU UBbtKVA 1 lUNb PUK 5HAPIRO-WILKS EXAMPLE
H Test Concentration (ppb)
• Replicate
H
m i
• 2
m 3
B 4
• 5
• 6
• 1
m 8
Control
0.13
0.13
-0.48
0.13
0.13
-0.32
0.13
0.13
50.0
0,07
-0.10
0.25
-
-0.03
0.07
-0.19
-0.10
100.0
0.09
0,26
-0.52
0.26
0.44
-0.52
0.00
~
  14.4.6,2  Calculate  the  denominator,  D,  of  the  statistic:
                            n
                        D  =  z  (Xi  - X)2
     Where    X-j =  the  ith  centered observation
                                   of the centered observations
             n  = the total number of centered observations
  14,4.6.3  For this  set of data:    n = 22
                                    X =   1   (0.000) = 0,
                                         22
                                    D -  1.4412
  14.4.6,4  Order the centered observations from smallest to largest
 where X(i) denotes the  ith ordered observation.  The ordered
 observations for this example are listed in Table 22,
                                     220

-------
     TABLE 22.  ORDERED CENTERED .OBSERVATIONS FOR SHAPIRO-WILKS EXAMPLE
i
T v" ^ '
2
3
4
5
6
7
8
9
10
11
X(i)
-0.52
-0.52
-0.48
-0.32
-0.19
-0.10
-0.10
0,03
0.00
0.07
0.07
i
12
13
14
15
16
17
18
19
20
21
22
x(D
0.09 ,-,-.;./.
0.13 '*£•"
0.13
0.13
0.13
0.13
0.13 v:;;:X -
0.25
0.26
0.26
0.44
14.4.6.5  From Table 4, Appendix B, for the number of observations, n,
obtain the coefficients a], 32, ... aj< where k is approximately
n/2. For the data in this example, n = 22 and k = 11.  The ai values
are listed in Table 23.

14.4.6.6  Compute the test statistic, W, as follows:
               W = 1 [  E a-f  (x(n-i-M)  -
The differences
in this example:
                         - X<*)  are listed in Table 23.   For the data
                  ^(1.1389)2=0.900
                                   22]

-------
    TABLE 23.  COEFFICIENTS AND DIFFERENCES FOR SHAPIRO-WILKS EXAMPLE
X
- x(D
- X(2)
- X(3)
- X(4)
- X(5)
- X(6)
- X(7)
- X(8)
- xO)
- xOo)
- xdi)
 14.4.6.7 The decision rule for this test is to compare W as calculated in
 14.4.6.6 to a critical value found in Table 6, Appendix B.  If the
 computed W is less than the critical value, conclude that the data are
 not normally distributed. For this set of data, the critical value at a
 signficance level of 0.01 and n = 22 observations is 0.900. Since W =
 0.897 is greater than the critical value, it is reasonable to assume that
 the data are normally distributed.

 14.4.7  Test for Homogeneity of Variance

 14.4.7.1  The test used to examine whether the variation in proportion of
female mysids with eggs is the same across all concentration levels
 including the control, is Bartlett's Test (Snedecor and Cochran,  1980).
The test statistic is as follows:
           B =
[  (  I Vj)  In
   1=1
                                      In
1=1
    Where:  V-j  =  degrees of freedom for each copper concen-
                 tration and control,  V-j = (nj  -  1)

           p  =  number of concentration levels including  the control
                                    222

-------
            S2 =
                       P
                       Z V1
                      1=1
           C  = 1 + ( 3(p-l))-l [ L 1/Vj - ( I ViH ]
                                 1=1        1=1

           In = loge

           i  = 1, 2, ..., p where p is the number of concentrations
                             including the control                i;;,c
           n-} = the number of replicates for concentration i.

  4.4.7.2  For the data in this example, (See Table 20} n] = 8, r\z = 7
 ind ns = 7,  Thus, the respective degrees of freedom are 7, 6 and 6.

 14.4.7.3  Bartlett's statistic is therefore:
        B =  [(19)ln(0.077) - (7 ln(0.064) + 6 ln(0.021) 4 6 ln(0.147))]/l.07

          *  [19(-2.564)  - (-53.925JJ/1.07

          *  [-48.716 - (-53.925)]/1.07   -/-:;,

          =  4.868

fl4.4.7.4  B is approximately distributed as chi-square with p - 1 degrees of
Ifreedom,  when  the variances are in fact the same.   Therefore, the
 ippropriate critical value for  this test, at a significance level of 0.01
 fith  two degrees  of  freedom,  is 9.210.  Since B =  4.868 is less than the
 critical  value of 9.210,  conclude  that the variances are not different.

H4.4.8  Bonferroni's T-test

114.4.8.1   Bonferroni's T-test is used  as an alternative to Dunnett's
 'rocedure when, as in this set  of  data,  the number of replicates is not  the
fsame  for all concentrations.   Like Dunnett's Procedure, it uses a pooled
[estimate  of the variance,  which is equal to the error value calculated  in an
^analysis  of variance. To obtain an estimate of the pooled variance,
fconstruct an ANOVA table  as described  in Table 24.
                                      223

-------
                           Table 24.  ANOVA TABLE
    Source
       df
    Total
      N - 1
                            Sum of Squares
                                 (SS)
                                 SST
Mean Square(MS)
    (SS/df)

Between

Within

P

N

- 1

- P

SSB

SSW
2
SB =
2
sw =

SSB/(p-1)

SSW/(N-p)
Where:
 p
 N
 n-j
               =  number  concentration  levels including the control
               =  total  number of observations  n]  + r\2 •••  +np
               =  number  of observations  in  concentration i
      P
SSB = Z
                                          Between Sum of Squares
SST = I   iVfV
     1=1 j=l

SSW = SST - SSB
                            - G2/N
                                          Total  Sum of  Squares


                                          Within Sum of Squares
                                                                 i
            G  = the grand total  of all  sample observations, G = I T-j
                                                                i = 1
            T-j  = the total of the replicate measurements  for
                 concentration "i"
           Yij  = the jth  observation for concentration "i"  (represents
                 the proportion of females with eggs for  concentration
                 i  in test chamber j)

14.4.8,2  For the data in this example:

    n]  - 8  n2  = 7   n3 =  7
    N  « 22
    T]  = YH +  YIZ  + ...  + YIB =  11.5                          ..- :
    T2  = Y2] +  Y22  + ...  + Y27 =   4.94                        ;v
    T3  = Y31 +  Y32  + .-  + Y37 =   3.65

    G  = TI 4 T2 +  TS + T4 = 20.09
                                    224

-------
     SSB = z T^/ni - G2/N
            132.25 + 24.40 + 13.32  -  403,6
SST =
P   ri
I   Z
                       7
                      - Q2/N

         = 23.396 - 403.61  = 5.05
                     22

     SSW = SST - SSB = 5.05 - 3.57 = 1,48

     SB2 = SSB/p-1 .= 3.57/3-1 = 1.785

     SW2 = sSW/N-p  = 1.48/22-3 = 0.078


 14.4.8.3  Summarize these calculations  in  the  ANOVA  table  (Table  25)

           Table 25.  ANOVA TABLE FOR DUNNETT'S PROCEDURE EXAMPLE
Source
Between
Within
Total
df
2
19
21
Sum of Squares
(SS)
3.57
1.48
5.05
Mean Square(MS)
(SS/df)
1.785
0.078

14.4.8.4  To perform the individual comparisons, calculate the t
statistic for each concentration, and control combination as follows
Where
      Sw

      nl
                                  »- d/nj)

 mean proportion of females with eggs  for concentration i
 mean proportion of females with eggs  for the  control
 square root of within  mean square
 number of replicates for  control
 number of replicates for  concentration  i.

                        22S

-------
 14.4.8.5  Table 26 includes the calculated t values for each
 concentration  and control combination.  In this example, comparing the
 50.0 ppb concentration with the control the calculation is as follows:

                                     { 1.44 - 0.52 )
                                 [ 0.279V U/8J + (1/7) ]
                        TABLE  26.   CALCULATED T-VALUES
              Test Concentration (ppb)
50.0
100.0
2
3
5.05
6.37
 14.4.8.6  Since the purpose of this test is to detect a  significant
 reduction in mean proportion of females with eggs,  a (one-sided)  test is
 appropriate.  The critical  value for this one-sided test is  found in
 Table 5, Appendix E,  Critical Values for Bonferroni's "T".   For  an
 overall  alpha level of 0.05, 19 degrees of freedom  for error and  two
 concentrations (excluding the control)  the approximate critical  value is
 2.094.   The mean proportion for concentration "1"  is considered
 significantly less than the raean proportion for the control  if t-; is
 greater  than the critical value.  Therefore,  the 50.0 ppb and the
 100.0 ppb concentrations have significantly lower mean proportion of
 females  with eggs than the  control.   Hence the LOEC for  fecundity is 50.0
 ppb.

 14.4.8.7  To quantify the sensitivity of the  test,  the minimum
 significant difference (MSD) that can be detected statistically may be
 calculated.
jWhere:  t
        Sw
        n  -
           = t Sw v7 (l/ni) + (1/n)

the critical value for Bonferroni's t-test
the square root of the within mean square
the common number of replicates at each concentration
(this assumes equal replication at each concentration
the number of replicates in the control.
 14.4.8.8   In this example:
                   MSD = 2.094  (0.279) / (1/8) + (1/7)
                       = 2.094  (0.279)(0.518)
                       = 0.303
                                    226

-------
 14.4.8.9  Therefore,  for this  set  of  data,  the  minimum  difference  that
Kan  be  detected as statistically significant  is 0.30.
E
fl4.4.8.10  The  MSD (0.30)  is  in  transformed units.   To  determine the  MSD
|in terms of percent of females with eggs, carry out  the following
 inversion.

§4.4.8.10.1   Subtract the  MSD  from the  transformed control mean.

                             1.44 - 0.30 = 1.14

14.4.8.10.2 Obtain the untransformed  values for the  control  mean and
 ;he  difference  calculated  in 4.10.1.

                           [Sine  (1.44)  ]Z   =  0.983
                           [Sine  (1.14)  ]Z = 0.823

[14.4.8.10.3  The untransformed MSD (MSDU) is  determined by subtracting
[the  untransformed values from  14.4.8.10.2.

                         MSDU = 0.983  -  0.823  =  0.16

[14.4.8.11  Therefore, for  this set of data, the minimum difference in
pean proportion of females with  eggs  between  the control  and any copper
fconcentration that can be  detected as statistically  significant  is 0.16.

 14.4.8.12  This represents a  17% decrease in  proportion of females with
 eggs from the control.
 14.4.9  Probit Analysis

;14.4.9.1   The data used for the probit  analysis is summarized in
[Table 27.   For the probit analysis,  the test concentration  with  0%
ifemales with eggs in all eight  replicates was considered.   To perform the
 probit analysis,  run the EPA Probit  Analysis Program,  using the  option to
[adjust for response in the controls.  An example of the program  out is
'provided in Table 28 and Figure 13.

 14.4.9.2  For this example, the chi-square test for heterogeneity was not
 significant.  Thus probit analysis appears to be appropriate for this set
 of data.
                                     227

-------
                    TABLE 27.  DATA FOR PROBIT ANALYSIS
                                     Test Concentration (ppb)
                        Control    50.0       100.0      210.0
No. Females W/0 Eggs       2
No. Females               19
13
22
10
16
14
14
                                    228

-------
    TABLE 28.   OUTPUT FROM  EPA PROBIT ANALYSIS PROGRAM,  VERSION  1.4.
                MYSID FECUNDITY DATA.
 Mysidopsis  Bahia Fecundity Data Analysis
     Cone.

    Control
    50.0000
   100.0000
   210.0000
 Number
Exposed

    19
    22
    16
    14
   Number
   Resp.

       2
      13
      10
      14
 Observed
Proportion
Responding

  0.1053
  0.5909
  0.6250
  1.0000
 Adjusted
Proportion
Responding

  0.0000
  0.5411
  0.5793
  1.0000
Predicted
Proportion
Responding

  0.1086
  0.4695
  0.7454
  0.9263
 Chi - Square Heterogeneity =     3.343
Mu
Sigma

Parameter
    1.730247
    0.408592

    Estimate
                            std. Err.
                                  95%  Confidence Limits
 Intercept
 slope

 Spontaneous
 Response Rate
    0.765340
    2.447431

    0.108593
      1.796586
      0.937746

      0.071352
           -2.755967,
            0.609448,

           -0.031257,
            4.286648)
            4.285414)

            0.248444)
      Estimated EC Values and Confidence Limits
Point

EC 1.00
EC 5.00
EC10.00
EC15.00
EC50.00
EC85.00
EC90.00
EC95.00
EC99.00
       Cone.
                Lower       Upper
              95% Confidence Limits
        6.
       11,
       16,
       20.
       53.
      142.
      179.
      252.
      479.
0220
4328
0915
2667
7337
4661
4314
5455
4620
    0.0022
    0.0289
    0.1127
    0.2814
   12.0495
   92,5911
 111.0103
 140.4689
 209.9131
     19.2441
     28.1558
     34,6456
     39.9812
     81.7779
   932.0419
  2239.7014
  8489.1182
107515.0700
                                     229

-------
Probit
   10+
    s+
    8+
    7+
    6+
    4 +
    3 +
    2-
     1+
     0+
       EC01
                     EC10     EC25      EC50      EC75     EC90
                                                                        EC99
Figure  13.   Plot  of adjusted probits and predicted  regression line
                                 230

-------
 :5.  PRECISION AND ACCURACY

   i  PRECISION

 -5.1.1  Data on the  single laboratory precision of the mysid survival,
Browth, and fecundity using copper sulfate  (CU) and  sodium dodecyl
 ulfate (SDS) in  natural  seawater are shown  in Tables 29 and 30.
 urvival NOEC/LOEC pairs  showed good precision, and  were the same  in four
   the six tests  with CU  and SDS.  Growth and fecundity were generally
 ot acceptable end points in either sets of  tests.

 [5.1.1  The multi-laboratory precision of the test has not yet been
 Fetermined.
 5.2  ACCURACY
 .5.2.1  The accuracy of toxicity tests cannot  be  determined.
                                     231

-------
TABLE 29. SINGLE LABORATORY PRECISION OF THE MYSID (MYSIDQPSIS BAHIA)
          SURVIVAL, GROWTH, AND FECUNDITY TEST PERFORMED IN NATURAL
          SEA WATER, USING JUVENILES FROM HYSIDS CULTURED AND SPAWNED  IN
          NATURAL SEAWATER, AND COPPER (CU) AS A REFERENCE
          TOXICANTl.2 34567
"
Survival

i Test
L
1 ,
r 2
1 3
«
1 5
1 6
NOEC
(ug/L)
63
(5)
125
125
125
125
LOEC
(ug/L)
125
(5)
250
250
250
250
Growth
NOEC
(ug/L)
(6)
(6)
(6)
(6)
(6)
(6)
LOEC
(ug/L)
(6)
(6)
(6)
(6)
(6)
(6)
Reproduction
NOEC
(ug/L)
(6)
(6)
(6)
(5)
(6)
125
LOEC
(ug/L)
(6)
(6)
(6)
(5)
(6)
(7)
Most
Sensitive
End Point
S
s
S
s
s

''Tests performed by Randy Cameleo, Environmental Research Laboratory,
 U. S. Environmental Protection Agency, Narragansett, Rhode Island.

2Eight replicate exposure chambers, each with five juveniles, were
 used for the control and each toxicant concentration.  The temperature
 of the test solutions was maintained at 26 + 1°C.

3Copper concentrations in Tests 1-2 were: 8, 16, 31, 63, and 125 ug/L.
 Copper concentrations in Tests 3-6 were, 16, 31, 63, 125, and  250 ug/L.

4For a discussion  of the precision of data from chronic toxicity
 tests see Section 4, Quality Assurance.

5Test results  inconclusive.

6No effect.
7SE = Survival  effects.  Fecundity data at these toxicant concentrations
 were disregarded  because there was a significant reduction in  survival.
                                       232

-------
STABLE  30.  SINGLE  LABORATORY  PRECISION  OF  THE MYSID  (MYSIDOPSIS  BAH-IA)
           SURVIVAL,  GROWTH,  AND  FECUNDITY TEST PERFORMED  IN NATURAL
           SEA WATER,  USING JUVENILES FROM MYSIDS CULTURED AND SPAWNED  IN
           NATURAL SEAWATER,  AND  SODIUM DODECYL SULFATE  (SDS) AS A
           REFERENCE  TOXICANTl,2  3  4 5  6 7

i Survival

;Test
i
1
1
! 2
f 3
4
5
6
NOEC
(mg/L)
2.5
(6)
(6)
5,0
2.5
5.0
LOEC
(mg/L)
5.0
(6)
(6)
10.0
5.0
10,0
Growth
NOEC
(mg/L)
(6)
(6}
(6)
(6)
(6)
(6)
LOEC
(mg/L)
(6)
(6)
(6)
(6)
(6)
(6)
Reproduction
NOEC
(mg/L)
(6)
(6)
(6)
(6)
(6)
5.0
LOEC
(mg/L)
(6)
(6)
(6)
(6)
(6)
(7)
Most
Sensitive
End Point
S
-
_
S
S
-
^Tests performed by Randy Cameleo, Environmental Research Laboratory,
 U. S. Environmental Protection Agency, Narragansett, Rhode Island.
       replicate exposure chambers, each with five juveniles, were
 used for the control and each toxicant concentration.  The temperature
 of the test solutions was maintained at 26 +
     concentrations in Tests 1-2 were: 0.3, 0.6, 1.3, 2.5, and
 5.0 mg/L.  SDS concentrations in Tests 3-4 were: 0.6, 1.3, 2.5, 5.0 and
 10.0 mg/L,  SDS concentrations in Tests 5-6 were: 1.3, 2.5, 5.0, 10.0,
 and 20.0 mg/L.
     a discussion of the precision of data from chronic toxicity
 tests see Section 4, Quality Assurance.

 }Test results inconclusive.

 'No effect.

 SE = Survival effects.  Growth data at these toxicant concentrations
 were disregarded because there was a significant reduction in survival.
                                      233

-------
                    f-'igure 14.  Data sheet for water quality measurements
                               From Lussler, Kuhn, and Sewall, 1987.
TEST:
START DATE;

SALINITY:

DAY 1

DAv 2

DAY 3

DAY 4

DAY 5

DAY 6

^AY 7



DAY 1

DAY ?

DAY 3

DAY 4

DAY 5

DAY 6

DAY 7


TRTMT
RFP
RFP
qc-p
HEP
REP
REP
REP
PEP
REP
REP
REP
REP
3eE
REP

TRTMT
REP
REP
PEP
RFP
REP
REP
REP
REP
REP
REP
-HEP
REP
REP
REP

TEMP















rt'Mf















SALINITY















SALINITY















DO















DO















pH















pH















TRTMT















T R 1 M r















TEMP















TEMP















SALINITY















SAL IN'"-V















• DO















I) O















pH















pH















































                                               234.

-------
              Figure 15.  Data  sheet for survival  and fecundity data
                         From  Ussier, Kuhn, and  Sewall, 1987.
TEST:
START DATE:

SALINITY:
TREATMENT/
REPLICATE
1
2
3
4
	
5
6
7
8
1
2
3
4
1 	
5
6
7
8
1
2
3
4
2 	
f
6
7
8
DAY 1
f ALIVE
























DAY 2
t ALIVE
























DAYS
# ALIVE






















-

DAY 4
1 ALIVE
























DAYS
if ALIVE
























DAY 6
* ALIVE
























DAY?
/ALIVE
























FEMALES
W/EGGS
























FEMALES
NO EGGS
























MALES
























MMATURES
•^^^•MHi
























                                          235

-------
                               Figure 15. Continued.
TEST:
START DATE:

I
iSALINITY:
TREATMENT/
REPLICATE
1
2
3
4
3
5
6
7
8
1
2
3
4
5
6
7
8
1
2
3
4
5
5
6
7
8
DAY 1
* ALIVE
























DAY 2
* ALIVE
























DAY 3
# ALIVE




















•



DAY 4
* ALIVE
























DAYS
* ALIVE
























DAY 6
t ALIVE
























DAY?
* ALIVE
























FEMALES
W/EGGS
























FEMALES
NO EGGS
























MALES
























iMMATURES
























                                        236

-------
               Figure 16.  Data  sheet for dry weight measurements
                           From  Lussier, Kuhn, and Sewall, 1987.
 *EST:
START DATE:_

 SALINITY:
TREATMENT
; REPLICATE

2
3
4
C 	
5
6
7
8
1
2
3
4
1 	

6
7
8
1
2
3
4
2 	
5
6
7
8
PAN*
























TARE
WT.
























TOTAL
WT.
























ANIMAL
WT.
























#OF
ANIMALS
























XWT./
ANIMAL
























                                     237

-------
                     Figure 16.  Continued.
iTEST:
 JTART DATE:_
[SALINITY:
TREATMENT
REPLICATE
1
2
3
4
3 	
5
6
7
8
1
2
3
4
4 	
5
6
7
8
1
2
3
4
5 	

6
7
8
PAN*
























TARE
WT.
























TOTAL
WT.
























ANIMAL
WT.
























#OF-
ANIMALS
























XWT./
ANIMAL
























                                   238

-------
                                   SECTION  15

                                 TEST METHODl.2

               SEA  URCHIN  (ARBACIA PUNCTULATA) FERTILIZATION TEST
                                   METHOD 1008
    SCOPE AND APPLICATION
 _.l  This method  measures  the  toxicity  of  effluents  and receiving water to  the
 jametes of the  sea  urchin, Arbac i a  punctulata,  during  a 1  h and  20 min
 jxposure.  The  purpose  of  the  sperm cell toxicity  test is  to determine the
 Concentration of  a  test substance that  reduces  fertilization of  exposed
gametes relative  to that of  the control.

>1.2  Detection  limits of the toxicity of an  effluent or pure substance are
[organism dependent.

|1.3  Single  or  multiple excursions  in toxicity  may not be  detected using  24-h
^composite samples.   Also,  because of the long sample collection  period
'involved in  composite sampling and  because the  test  chambers are not  sealed,
^highly volatile and highly degradable toxicants in the source  may not be
'detected in  the test.

 1.4  This method  should be restricted to use by, or  under  the  supervision of,
 professionals experienced  in aquatic toxicity testing.

 2.   SUMMARY  OF  METHOD

t2.1  The method consists of  exposing dilute  sperm  suspensions  to effluents  or
'receiving waters  for one hour.  Eggs are then added  to the sperm suspensions.
Jwenty minutes  after the eggs  are added, the test  is terminated  by the
 addition of  preservative.  The percent  fertilization is determined by
•microscopic  examination of an  aliquot from each treatment.  The  test  results
'are  reported as the concentration of the test substance which  causes  a
^statistically significant  reduction in  fertilization,  compared to the
'controls.

[3.   DEFINITIONS

     (Reserved  for  addition  of terms at a  later date).

     INTERFERENCES

 4.1  Toxic substances may  be introduced by contaminants in dilution water,
 glassware, sample hardware,  and  testing equipment  (see Section 5, Facilities
 and  Equipment).
      format  used for this method  was  taken from Kopp,  1983.
 2This method was adapted from Nacci,  Walsh,  and Jackim,  1987,  Environmental
 Research  Laboratory, U.  S. Environmental  Protection Agency,  Narragansett,
 Rhode Island.
                                       239

-------
[4.2   Improper effluent sampling and handling  may adversely affect test
'results  (see  Section 8,  Effluent and Receiving Water Sampling and Sample
 Handling).

     SAFETY

15.1   See Section  3,  Health and Safety.

     APPARATUS AND EQUIPMEiNT

  .l  Facilities for holding and acclimating test organisms.

      Laboratory Arbacia  punctulata culture unit — See cuHuring methods
 ielow.   To  test effluent or receiving water toxicity,  sufficient eggs and
Sperm must  be available.

[-6.3   Samplers --  automatic sampler, preferrably with sample cooling
Icapability, that  can collect a 24-h composite sample of 1  L.

;6.4   Environmental chamber or equivalent facility with temperature control
 (20  +_ 1°C)  for controlling temperature  during exposure.

•6.5   Water  purification  system -- Millipore Super-Q, Deionized water (DI) or
^equivalent.
i
 6.6   Balance  -- Analytical, capable of  accurately weighing to 0.0001 g.

,6.7   Reference weights,  Class S -- for  checking performance of balance.

 6.8   Air pump --  for supplying air.
i
'6.9   Air lines, and  air  stones -- for aerating water containing adults.

,6.10 Vacuum  suction device -- for washing eggs.
i
i
;6.11 pH and  DO meters -- for routine physical and chemical measurements.
jUnless the  test is being conducted to specifically measure the effect of one
[of these two  parameters, portable, field-grade instruments are acceptable.

 6.12 Standard or micro-Winkler apparatus —  for  determining DO (optional).
 L
 6.13 Transformer, 12 Volt, with steel  electrodes  --  for  stimulating
 release  of  eggs and  sperm.

^6.14 Centrifuge, bench-top, slant-head, variable speed  -- for washing  eggs.
f
;6.15 Fume  hood -- to protect the analyst from formaldehyde fumes.

;6.16 Dissecting  microscope -- for counting diluted egg stock.

:6.17 Compound microscope — for examining and counting sperm cells and
|fertilized  eggs.

                                      240

-------
  .18  Sedgwick-Rafter counting chamber -- for counting egg stock.
  ..19  Hemacytometer, Neubauer -- for counting sperm.
06.20  Count register, 2-place -- for recording sperm and egg counts.
|6.21  Refractometer -- for determining salinity.
1
|p.22  Thermometers, glass or electronic, laboratory grade -- for measuring
f|/ater temperatures.
  i.23  Thermometers, bulb-thermograph or electronic-chart type -- for
 Continuously recording temperature.
  i.24  Thermometer, National Bureau of Standards Certified (see USEPA METHOD
|170.1, USEPA, 1979) — to calibrate laboratory thermometers.
16.25  Ice bucket, covered -- for maintaining live sperm.
j
|6.26  Centrifuge tubes, conical -- for washing eggs.
  ..27  Cylindrical glass vessel, 8-cm diameter -- for maintaining dispersed
 ;gg suspension.
15.28  Beakers -- six Class A, borosilicate glass or non-toxic plasticware,
iflOOO ml for making test solutions.
J
15.29  Glass dishes, flat bottomed, 20-cm diameter -- for suspending eggs.
i
p.30  Wash bottles -- for deionized water, for rinsing small glassware and
finstrument electrodes and probes.
  '.31  Volumetric flasks and graduated cylinders -- Class A, borosilicate
fglass or non-toxic plastic labware, 10-1000 ml for making test solutions.
  i.32  Syringes,  1-mL, and 10-mL, with 18 gauge, blunt-tipped needles (tips
 :ut off) -- for  collecting sperm and eggs.
|fe.33  Pipets, volumetric -- Class A, 1-100 ml.
  i.34  Pipets, automatic -- adjustable,  1-100 ml.
  ..35  Pipets, serological -- 1-10 mL, graduated.
§5.36  Pipet bulbs and fillers — PRQPIPETR, or equivalent.
I
[6.37  Tape, colored -- for labelling tubes.
I
[6.38  Markers, water-proof -- for marking containers,  etc.
                                       241

-------
 7.  REAGENTS AND CONSUMABLE MATERIALS

 7.1  Sea Urchins, Arbacia punctulata (approximately 12 of each sex).

 7.2  Food — kelp, Laminaria sp., or romaine lettuce for A, punctulata.

 7.3  Standard salt water aquarium or Instant Ocean Aquarium {capable of
 maintaining sea water at IB^C), with appropriate filtration and aeration
 system.

  .4  Sample containers — for sample shipment and storage (see Section 8,
 Affluent and Receiving Water Sampling and Sample Handling).

  '.5  Scintillation vials, 20 ml, disposable — to prepare test
 ^concentrations,

 |7.6  Parafilm — to cover tubes and vessels containing test materials.

 |7.7  Gloves, disposable — for personal protection from contamination.
 I
 7.8  Data sheets (one set per test) —  for data recording (see Figures  4, 5,
 [and 6).
 1
 .7.9  Acetic acid, 10%, reagent grade, in sea water --  for preparing killed
 |sperm dilutions.
 I
 7.10  Formalin, }Q%S  buffered (1,620 ml distilled water,  620 ml
 [formaldehyde,  reagent grade)9(6.48 g NaH2P04 or KH2PG4,  10,5 g
    P04 or K2HP04) --  for preserving eggs.

 .7.11  pH buffers 4,  79 and 10 (or as per instructions  of  instrument
 ^manufacturer)  for standards  and  calibration check (see USEPA Method 150.1,
 USEPA,  1979).

J7.12  Membranes and  filling  solutions for  dissolved  oxygen  probe  (see USEPA
jMethod  360.1,  USEPA,  1979),  or reagents for modified Winkler analysis.

|7.13  Laboratory quality assurance samples and  standards  for the  above
• methods.

 <7.14  Reference  toxicant  solutions (see Section  4, Quality  Assurance).
I
j7.15  Reagent  water -- defined as distilled or  deionized  water that does  not
^contain substances which  are  toxic to the  test  organisms.

 7.16 Effluent,  surface water, and dilution water — see  Section  7,  Dilution
 Water,  and  Section 8,  Effluent and Surface Water  Sampling and  Sample
 Handling.

 7.16.1   Saline test and dilution  water  --  The salinity of the  test  water
 must be 30 °/oo.  The  salinity should vary by no  more than + 2  o/oo
 among the replicates.
                                      242

-------
  7.16.2  The overwhelming majority of industrial and sewage treatment
  effluents entering marine and estuarine systems contain little or no
  measurable salts.   Exposure of sea urchin eggs and sperm to these effluents
  will  require adjustments in the salinity of the test solutions.   It is
  important to maintain a constant salinity across all treatments.   Two
  methods  are available to adjust salinities hypersaline brine derived from
  natural  seawater or artificial sea salts.   Use of hypersaline brine will
  limit  the concentration of effluent tested to  703.

  7.16.3  Hypersaline brine:   Hypersaline brine  (HSB)  has several  advantages
  that make it desirable for use in  toxicity testing.   It can be made from  any
  high quality,  filtered seawater fay evaporation,  and  can be  added  to the
 Affluent  or to deionized water to  increase the salinity.   HSB derived from
  atura!  seawater contains the  necessary trace  metals,  biogenic colloids,  and
 |Some of  the microbial  components necessary for adequate growth,  survival,
 |ind/or reproduction of marine  and  estuarine organisms,  and  may be  stored  for
 •"rolonged periods  without any  apparent  degradation.

 7.16.3.1   The  ideal  container  for  making HSB from natural seawater  is  one
 |that (1   has a  high  surface  to  volume ratio, (2)  is  made of  a  non-corrosive
 ^material,  and  (3)  is easily  cleaned  (fiberglass  containers  are ideal).
 jSpecial care should  be  used  to  prevent  any toxic  materials  from coming in
 Pcontact with the seawater being  used to generate  the brine.   If a heater  is
 hmmersed  directly  into  the seawater, ensure that  the heater  materials  do not
 .corrode or  leach any substances  that would  contaminate  the brine.   One
 ^successful method  used  is a  thermostatically controlled heat exchanger made
 pom fiberglass.   If aeration  is used, use only oil-free air compressors to
 Jprevent contamination.                        .

 p.16.3.2  Before adding seawater to the brine generator, thoroughly clean
 |the generator, aeration supply tube, heater, and any other materials that
 .will be in direct contact with the brine.  A good quality biodegradable
|detergent should be used, followed by several thorough deionized water
 rinses.  High quality (and preferably high salinity) seawater should be
 |Filtered to as least 10 urn before placing into the brine generator.  Water
^should be collected on an incoming  tide to minimize the possibility of
icontamination.
 1
 J.16.3.3  The temperature of the seawater is increased slowly to  40°C.
LThe water should be aerated to prevent temperature stratification and to
 increase  water evaporation.   The brine should be checked daily (depending  on
|the volume being generated)  to  ensure that the  salinity does not  exceed
1100 o/oo  and that the temperature does not exceed 4QQC.  Additional
^seawater  may be added to the  brine  to obtain the volume of  brine  required.

|.16.3.4   After the required  salinity is attained, the  HSB  should  be
pltered  a second time  through  a 1-um filter and poured directly  into
.portable  containers (20-L cubitainers  or polycarbonate  water cooler  jugs are
^suitable).   The containers should be capped and labelled with the  date the
i>rine was  generated and its salinity.  Containers of  HSB should be  stored  in
     dark  and maintained under room  temperature  until  used.
                                      243

-------
(7.16.3.5   If a source of HSB  is  available,  test  solutions  can  be  made  by
^following  the directions below.   Thoroughly mix  together the deionized water
 and brine  before mixing in the effluent.
I
 7.16.3.6   Divide the salinity of the  HSB  by the  expected test  salinity to
[determine  the proportion of deionized water to brine.   For example,  if the
|salinity of  the brine is 100  o/oo and the test is  to  be conducted at 30
|°/oo,  100  °/oo divided by 30  °/oo = 3.3.   The proportion of brine is
|1  part  in  3.3 (one part brine to 2.3  parts  deionized  water).

17.16.3.7   To make 1  L of seawater at  30 °/oo salinity from a hypersaline
 irine  of 100 °/oo, 300 ml of  brine and 700  ml of deionized water  are
Inquired.
I
^7.16.3.8   Table 1 illustrates the preparation of test solutions at 30  °/oo
.if they are  made by combining effluent (0 °/oo), deionized water  and HSB
 (100 °/oo),  or Forty Fathoms^ sea salts.

 7.16.4  Artificial sea salts: Forty  FathomsR brand sea salts  have been
 used successfully at the EMSL-Cincinnati  Newtown Facility  for  long-term
 (6 to  12 months) maintenance  of  stock cultures of  sexually mature sea
 urchins and  to perform the sea urchin fertilization test.

 7.17  SEA  URCHINS

 7.17.1  Adult sea urchins (Arbacia punctulata) can be obtained from
 commercial suppliers.  After  acquisition, the animals are  sexed by briefly
 stimulating  them with current from a  12 V transformer.  Electrical
 stimulation  causes the immediate release  of masses of gametes  that are
 readily identifiable by color -- the  eggs are red, and the sperm  are white.

 7.17.2  The  sexes are separated  and maintained in  20-L, aerated fiberglass
 tanks,  each  holding  about 20  adults.   The tanks  are supplied continuously
 (approximately 5 L/min) with  filtered natural seawater, or salt water
 prepared from commerical sea  salts is recirculated.   The animals  are checked
 daily  and  any obviously unhealthy animals are discarded.

 7.17.3  The  culture  unit should  be maintained at 15°C + 3°C, with a
 water  temperature control device.

 7.17.4  The  food consists of  kelp (Laminaria sp.), gathered from  known
 uncontaminated zones or obtained from commerical supply houses whose kelp
 comes  from known uncontaminated  areas, or romaine  lettuce. Fresh food is
 introduced into the  tanks at  approximately  one week intervals.  Decaying
 food is removed as necessary. Ample  supplies of food should always  be
 available  to the sea urchins.

 7.17.5  Natural or artificial seawater with a salinity of  30 °/oo is used
 to maintain  the adult animals, for all washing and dilution steps, and as
 the control  water in the tests (see Par.  7.16).
                                      244

-------
TABLE  1.
Control
0.0
                                       Solutions To Be Combined
•Effluent
JH;So1ution
I1
H 2
1 3
• 4
•
Effluent
Cone.
(%}
loo?
32
10
3.2
1.0
Volume of Volume of Diluent
Effluent Seawater (30 o/00)
Solution
600 ml
200 ml Solution 1 +
200 ml Solution 2 +
200 ml Solution 3 +
200 ml Solution 4 +
— __
400 ml
400 ml
400 ml
400 ml
                                                           400 ml
     Total
                                                          2000  ml
    his illustration assumes:  (1)  the  use  of  5  ml  of  test
   solution  in  each  of four  replicates {total of 20 ml)  for  the  control
   82 nf]veConcentrations of  effluent,  (2) an  effluent  dilution factor
   ot u.J,  (3)  the effluent  lacks  appreciable salinity,  and  (4)  400 ml of
   each test  concentration is  used for chemical analysis..   A sufficient
   initial volume  (600 ml) of  effluent is  prepared by adjusting  the
   salinity to  30 o/oo.   In  this example,  the salinity is adjusted by
   adding artificial  sea  salts  to  the  100% effluent,  and preparing a
   serial dilution using  30  o/00 seawater  {natural seawater,
   hypersaline  brine,  or  artificial seawater}.  Stir  solutions 1 h to
   ?nn*re«5at  !h? SaItS  d1ssolve-  The  salinity of the  initial  600 ml of
   100% effluent is  adjusted to 30 o/oo  by adding 18  g of dry
   artificial sea salts (Forty Fathoms*}.  Test concentrations are then
   made by mixing appropriate volumes  of salinity adjusted effluent and
   JO  o/oo salinity  dilution water to  provide 400 ml of solution for
   each  concentration.  If hypersaline brine alone (100 o/00) is used
   to  adjust the salinity of the effluent,-the highest concentration of
   effluent that could be tested would be  7Q% at 30 o/00 salinity
                                   245

-------
17.17.6   Adult  male and  female animals  used  in  field  studies  are  transported
 in separate  or partitioned insulated boxes  or  coolers  packed with  wet  kelp
'or paper toweling.  Upon arrival  at the  field  site,  aquaria  (or  a  single
partitioned  aquarium)  are filled  with  control  water,  loosely covered with a
'styrofoam sheet and allowed to equilibrate  to  15°C before  animals  are
padded.   Healthy animals will  attach to the  kelp or aquarium  within hours.

,-7.17.7   To successfully maintain  about 25  adult animals  for  seven  days at a
"field site,  a  screen-partitioned, 40-L glass  aquarium using  aerated,
^circulating, clean saline water {30  °/oo) and a gravel bed filtration
System,  is housed within a water  bath, such as an INSTANT  OCEAN* Aquarium
 (15°C).   The inner aquarium is used to avoid  contact  of  animals  and water
 iath with cooling coils.

  .  SAMPLE COLLECTION,  PRESERVATION AND HANDLING

 8.1  See Section 8, Effluent and  Receiving  water Sampling  and Sample
 Handling.

J9.  CALIBRATION AND STANDARDIZATION

 9.1  See Secion 4, Quality Assurance.

 10.  QUALITY CONTROL

 10.1  See Section 4, Quality Assurance.

 11.  TEST PROCEDURE

 11.1  TEST SOLUTIONS

 11.1.1   Surface Waters

 11.1.1.1  Surface water toxicity is determined with  samples  used directly as
 collected.

 11.1.2  Effluents

 11.1.2.1  The selection of the effluent test concentrations  should be  based
 on the objectives of the study.  One of two dilution factors, approximately
 0.3 or 0.5, is commonly used.  A dilution factor of  approximately 0.3  allows
 testing between  100% and 1% effluent using only five effluent concentrations
 ,100%, 30%, 10%,  3%, and 1%).  This series of dilutions minimizes  the  level
 of effort, but because of the wide  interval between  test concentrations
                                       246

-------
   rovides poor test precision.  A dilution factor of 0.5 provides greater
  precision, but requires several additional dilutions to span the same range
  .pf effluent concentrations.   Improvements in precision decline rapidly as
  the dilution factor is increased beyond 0.5.

  £1.1.2.2  If the effluent is known or suspected to be highly toxic,  a lower
  jange of effluent concentrations should be used (such as 10%,  3%,  1%, and
   . 1%).
  11.1.3  Control  Water
   1.1.3.1   Prepare 3 L of control  water  at  30 o/00  using  hypersaline  brine
 |r  artificial  sea salts  (see Table 1).   This water is  used  in  all  washing
 gnd diluting  steps and as control  water in the  test.   Natural  sea  water  and
 local  waters  may be used as  additional  controls.
  '11.1.4   Effluent  Dilutions
 m.1.4.1
 [30  o/oo.
Effluent/receiving water samples are adjusted to salinity of
 P
 ill.1.4.2  Four replicates  (minimum of  three) are prepared for each test
 Concentration, using  5 ml  of  solution  in disposable  liquid scintillation
 Ivials.  A 50% (0.5) concentration series can be prepared by serially
  liluting test concentrations  with control water.

 111.1.4,3  All test samples are equilibrated at 2Q°C +  1°C before
  addition of sperm.

 ill.2.   REFERENCE TOXICANT TEST

  11.2.1  A reference toxicant  test using copper sulfate is performed
  ;ide-by-side with each fertilization test, or set of tests, performed with a
  liven batch of gametes (see Section 4, Quality Assurance).

 |l.3.   COLLECTION OF GAMETES  FOR THE TEST

  1.3.1  Select four males and place in shallow, bowls, barely covering the
  nimals with seawater.  Stimulate the release of sperm by touching the shell
  ith steel electrodes connected to a 12 V transformer (about 30 seconds each
  ime).  Collect the sperm (about 0.5-1.0 ml) from each male,  using a 1-3 mL
  isposable syringe fitted with an 18-gauge,  blunt-tipped needle.  Pool the
 |perm.  Maintain the pooled sperm sample on ice.   The sperm must be used in
   toxicity test within 1  h of collection.

 |11.3.2  Select four females and place in shallow bowls, barely covering the
jshell with seawater.   Stimulate the  release  of eggs and collected with a
[needle as described above.   Remove the needle from the syringe before adding
 |the eggs to  a  conical  centrifuge tube.   Pool  the  eggs.   The egg  stock may be
  leld at room temperature  for several  hours before use.   Note:   The egg
 Isuspension may be prepared during the 1-h  sperm exposure.
                                      247

-------
11.4.  PREPARATION OF SPERM DILUTION FOR USE IN THE TEST

11.4.1  Using control water, dilute the pooled sperm sample to a
concentration of about 5 X 107 sperm/mL (SPM).  Estimate the sperm
concentration as described below:.

    1. Make a sperm dilutions of 1:50,  1:100,  1:200,  and 1:400,  using
       30 o/oo seawater, as follows:

       a. Add 200 uL of collected  sperm to  10  ml of sea  water  in Vial  A
          Mix by gentle pipetting  using a 5-mL pipetter.
       b. Add 5 ml of sperm suspension  from Vial  A  to 5  ml  of  seawater in
          Vial B.   Mix by gentle pipetting  using a  5-mL  pipetter.
       c. Add 5 mL of sperm suspension  from Vial  B  to 5  mL  of  seawater in
          Vial C.   Mix by gentle pipetting  using a  5-mL  pipetter.
       d. Add 5 mL of sperm suspension  from Vial  C  to 5  mL  of  seawater in
          Vial D.   Mix by gentle pipetting  using  a  5-mL  pipetter.
       e. Discard  5 mL from Vial D.  (The volume  of  all suspensions  is  5 mL)

    2.  Make  a 1:2000 killed sperm suspension and  determine  the SPM.
                                                          Cap Vial C and
Add 5 mL 10% acetic acid in seawater to Vial C.
mix by inversion.
Add 1 mL of killed sperm from Vial C to 4 ml of seawater in
Vial E.  Mix by gentle pipetting with a 4-mL pipetter.
Add sperm from Vial E to both sides of the Neubauer
hemacytometer.  Let the sperm settle 15 min.
Count the number of sperm in the central 400 squares on both sides
of the hemacytometer using a compound microscope (400X).
the counts from the two sides.
SPM in Vial E  =  10^ x average count.
                                                                   Averaae
   3. Calculate the SPM in all other suspensions using the SPM in Vial  E
      above:

         SPM in Vial A  =  40 x SPM in Vial E
         Spm in Vial B  =  20 x SPM in Vial E
         SPM in Vial D  =   5 x SPM in Vial E
         SPM in original sperm sample  =  2000 x SPM in Via)  E

   4. Dilute the sperm suspension with a SPM greater than  5 x  10? SPM to
      5 x 10'  SPM.

         Actual  SPM/(5 x 107}  = dilution factor  (DF)

         [(DF)  x 5]  -  5 =  mL  of seawater to add  to  vial.

   5.  Confirm  the sperm count  by sampling  from the  test  stock.  Add 0.1  mL
      °f-^e^  s?ock  to 9'9 mL  of 10%  acet1c ac1d  i'n seawater,  and  count
      with  the  hemacytometer.   The  count should  average  50+5
                                    248

-------
 11.5  PREPARATION OF EGG SUSPENSION FOR USED IN THE TEST

Jl 1.5.1  Wash the pooled eggs three times using control  water with gentle
 :entrifugation (500xg for 3 min using a tabletop centrifuge).   If the wash
 later becomes red,  the eggs have lysed and must be discarded.

  1.5.2  Dilute the  egg stock, using control water, to about 2000 eggs/mL.

     1. Add control  water to bring the eggs to a volume  of 200 ml ("egg stock")

     2. Mix the egg  stock using an air-bubbling device.   Using  a  wide-mouth
        pipet tip, transfer 1 mL of eggs from the egg stock to  a  v^ial
        containing 9 mL of control water.  (This vial contains  an egg
        suspension diluted 1:10 from egg stock),

     3. Mix the contents of the vial using gentle pipetting.  Using a  wide-
        mouth pipet  tip, transfer 1 ml of eggs from the  vial to a
        Sedgwick-Rafter counting chamber.  Count all eggs in the  chamber using
        a dissecting microscope at 10X ("egg count").

     4. Calculate the concentration of eggs in the stock.  Eggs/mL = 10X (egg
        count).   Dilute the egg stock to 2000 eggs/mL by the formula below.

        a.  If the egg count is equal to or less than 200:

          (egg count)  - 200 = volume (ml) of control water to add
          to egg stock

        b.  If the egg  count is less than 200,  allow the  eggs to settle and
           remove enough control water to concentrate the eggs  to greater
           than 200,  repeat the count,  and dilute the egg stock as in  #1
           above.

           NOTE:     It requires 24 mL of a egg stock solution for each test
                    with a control and five exposure concentrations.

        c.  Transfer  1  mL of the diluted egg stock to a vial  containing 9 mL
           of control  water.   Mix well,  then transfer 1  mL from the vial  to a
           Sedgwick-Rafter counting chamber.   Count all  eggs using a
           dissecting  microscope.   Confirm that the final egg count =  200/mL.

   .6   START OF THE  TEST

   .6.1   Within  1  h  of collection add 100 uL  of appropriately diluted  sperm
ito each  test vial.   Record the time of sperm  addition.
!ll.6.2   Incubate all test vials at 20 +
                                            for  1  h.
 11.6.3  Mix the diluted egg  suspension  (2000 eggs/mL),  using gentle
 bubbling.  Add 1 mL of diluted  egg  suspension to each test vial using  a wide
 mouth pipet tip.   Incubate 20 min at  20 +
                                      249

-------
  11.7  TERMINATION OF THE TEST

 .11.7.1  Terminate the test and preserve the samples by adding 2 mL of
 •buffered formalin to each vial.

 til. 7. 2  Vials may be evaluated immediately or capped and stored for as lonq
     one week before being evaluated.

 £1.7.3  To determine fertilization,  transfer about 1 ml  eggs  from the bottom
 pt_a test vial  to a Sedgwick-Rafter  counting chamber.  Observe the eqqs
  jsing  a compound microscope (100  X).   Count between TOO  and 200
 Leggs/sample.   Record the number counted and the  number unfertilized
  ertilization is indicated by the presence of a  fertilization  membrane
  urrounding the egg.   Note:  adjustment of  the microscope to obtain proper
  ontrast  may be required to  observe  the fertilization membrane.   Because
 Camples are fixed in  formalin, a  ventilation hood  is set-up surrounding the
 microscope to protect the analyst from prolonged exposure  to formaldehyde
 •f
    .  ACCEPTABILITY OF TEST RESULTS
           nK!rmieg? ratl'° routine]y employed should result in fertilization
 of 70% to 90,4 of the eggs in the control chambers.

 |13.  SUMMARY OF TEST CONDITIONS
 i     —          •       —

 ,13.1  A summary of test conditions is listed in Table 2.

 J14.  DATA ANALYSIS

 •14.1  General                                                      ...
 i
 .14.1.1  Tabulate and summarize the data.  Calculate the percent of
 (unfertilized eggs for each replicate.   A sample set of test data is listed
    Table 3.
 i
 ;]4.1.2  The  endpoints of toxicity tests using the sea urchin  are based on
 ^rrc rr^ntl0n in Percent of e99s fertilized.   Point estimates,  such as ECU
 ;LLb, EC10 and EC50, are calculated using Probit Analysis (Finney,  1971)   A
 .hypothesis test  approach such as Dunnett's  Procedure (Dunnett,  1955} or
      lS^M,a^°neiRank Tcst (Steel>  1959; Miller>  1981)»  *s "sed  to estimate
      and LOEC values.   See the  appendices for examples of the manual
 computations, program listings,  and examples  of data input and  program
;output.                                                            3

,14.1.3   Formal statistical  analysis of  the  fertilization data is outlined in
 Mgure  1.  The response used  in  the analysis  is the proportion of
                                      250

-------
     1-   Test type

    2-  Salinity:

    3-  Temperature

    4-   Light quality:


   5-  Light intensity:
   6.   Test vessel  si
                 ze
  7-  Test solution volume:

  8.  Number of sea urchins
  9.


10.


11.
 Number of replicate
   chambers per treatment:

 Dilution  water:
    Dilution factor:
Test durat
              ion
   Effects
           measured:
   Number of treatments
     per test:
                                       Static

                                       30 o/00 ±

                                       20
                                      Ambient laboratory light
                                      during test preparation;

                                      10-20  u£/m2/Sa or 50-100 ft r
                                      (AmfaTent  laboratory levels)
                                     5 mL
           are  used per test

        2S.1S ,'?,a,s-M°'™»
    (minimum of 3}

  Uncontaminated source of -
 0-3 or 0.5

 1  n and  20 min

Fertilization of sea urchfn


Minimum of five  effluent
concentrations and a  control
                               251

-------
                TABLE 3.  DATA FROM  SEA  URCHIN FERTILIZATION TESTl
                	—   	
  Copper        Replicate
Concentration
  (ug/L)
No.  of Eggs    No.  of Eggs
  Counted       Unfertilized
  Percent
Unfertilized
K 	 ~- — 	
1 ° A
I n
' B
°
2.5 : A
n
B
C
5.0 A
B
C
10.0 A
B
C
20.0 A
B
C
40.0 A
B
C
	 — 	 	 	
	 	 	 	 — , —
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
— — ' '
15
22
13
19
35
29
37
26
22
37
34
49
59
59
63
ftfl
oo
70
74
~ 	 — 	
15
22
13
19
35
29
37
26
22
37
34
49
59
59
63

88
70
74
                             MnnJeC.nical APP^"cat1ons  Inc.,
                             Momtonng and Support  Laboratory
                                 252

-------

  -^ss;i;-!',rsrs,^«;, IK
informat10n on the Bonferroni adjustment see Appendix"!
;
                                          !
eggs compared to the control. In this analysis. tK?centaqesSf
unfertile eggs for all replicates at a given rtncentmK" combined.
                f^^^irn^ -s ss.t.srjrr of
           recommended for analysts who are not proficient in statistics.
                          253

-------
 14.2  EXAMPLE OF ANALYSIS OF SEA URCHIN FERTILIZATION DATA.

 114.2.1  This example uses toxicity data from a sea urchin (Arbacla
 mnctulata) fertilization test performed with copper.  The response of
 pnterest is the proportion of unfertilized eggs,  thus each replicate must
 cirst be transformed by the arcsin transformation procedure described in
 Appendix B.  The raw and transformed data, means  and standard deviations of
 ;he transformed observations at each copper concentration and control are
 'isted in Table 4.  The data are plotted in Figure 2.
                    TABLE  4.  SEA URCHIN FERTILIZATION DATA
                                        Copper Concentration (ug/L)
Replicate     Control
                                    2,5
5,0
10.0
20.0
40.0
m
mm A
IH RAW B
I
•
• ARC SINE A
: M TRANS- B
H FORMED C

K> j
Ssfc
M 	
0.15
0.22
0.13
0.398
0.488
0.369
0,418
0,004
1
0.19
0.35
0.29
0.451
0.633
0.569
0.551
0.008
2
0.37
0.26
0.22
0.654
0.535
0.488
0.559
0.007
3
0.37
0.34
0.49
0.654
0.622
0.775
0.684
0.001
4
0.59
0.59
0.63
0.876
0.876
0.917
0.890
0.0010
5
0.88
0.70
0.74
1.217
0.991
1.036
1.081
0.014
6
14.2.2   Test  for Normality

 14,2.2.1   The first  step of  the  test  for  normality is  to center  the
 observations  by subtracting  the  mean  of all  observations within  a
^concentration from each  observation in that  concentration. The centered
Observations  are summarized  in Table  5.
                                     254

-------
STATISTICAL ANALYSIS OF SEA URCHIN FERTILIZATION TEST
FERTILIZATION DATA
PERCENT UNFERTILIZED
1
* 1
PROBIT ARCSIN
ANALYSIS TRANSFORMATION
j j
ENDPOINT FSTTMiTF ^HArinn WT! K" TI^T ^ N°
EC1 EC5 EC10 EC50

NORMAL DISTRIBUTION!
| 	 BARTLETT'S TEST 	 M
HOMOGENEOUS VARIANCE
i t
^ 	 . EQUAL NUMBER OF EQUAL NUMBER C
REPLICATES? REPLICATES?
YES YES
v 1 i 	
InII|npnuTH DUNNETT'S STEEL'S MANY-ONE WIL
BuNi-tHHUNj, TF=;T RANK TFQT
ADJUSTMFNT ' tb f MANK ^tbl BOMFE


f
ENDPOINT ESTIMATES
NOEC, LOEC

RMAL DISTRIBUTION
HETEROGENEOUS
VARIANCE
F 	 ^S
lr
:OXON RANK SUM
TEST WITH
RRONI ADJUSTMENT




Figure 1.  Flow chart for statistical analysis of Arbacia data



                              255

-------
                                                          Dl
                                                          o»
                                                          QJ
                                                          O)
                                                          N
                                                          01
                                                    u
                                                   §
                                                   u
                                                   o
                                                   o
                                                   K
                                                   id
                                                          QJ
                                                          OJ
S333
            256

-------
          TABLE -5.  CENTERED OBSERVATIONS FOR SHAPRIO-WILKS EXAMPLE
i Copper Concentration
I Replicate Control
1
i
1 A -0.
I B 0.
1 ° ~°*
020
070
049
2.
-0.
0.
0.
5
100
082
018
5.
0.
-0.
-0.
0
095
024
071
10.
-0.
-0.
0.
0
030
062
091
20
-0
-0
0
Cug/L)
.0
.014
.014
.027
40
0
-0
-0
.0
.136
.090
.045
[14.2.2,2    Calculate the denominator, D, of the statistic:


                             (Xi - X)2


     Where   X-\ = the ith centered observation
             X  = the overall mean of the centered observations
             n  = the total number of centered observations

 14.2.2.3  For this set of data,    n = 18
                                    X = J_ (0) = 0
                                         18
                                    D = 0.0822

 14.2.2.4  Order the centered observations from smallest to largest

                X(l) - x<2> - ...  - x(n)

Iwhere x(i)  denotes the ith ordered observation.   The ordered
^observations  for this example  are listed in  Table 6.

   TABLE  6.   ORDERED CENTERED OBSERVATIONS FOR SHAPIRO-WILKS EXAMPLE
!
|
1
2
3
4
I 5
&
1 7
8
9
xH)
-0.100
-0.090
-0.071
-0.062
-0.049
-0.045
-0.030
-0.024
-0.020
i
10
n
12
13
14
15
16
17
18
xM>
-0.014
-0,014
0.018
0.027
0.070
0.082
0.091
0.095
0.136
                                    257

-------
 14.2.2.5  From Table 4, Appendix B, for the number of observations, n,
 obtain the coefficients a]," 32, ... a^ where k is approximately
 n/2. For the data in this example, n = 18 and k = 9.   The ai values are
 listed in Table 7.

 15.2.2.6  Compute the test statistic, W,  as follows:

                    i   k
                W = 1 [ 2 ai  (x(n-i+D - X

- X(3)
- X<4)

- X(6)

- X(8)

14.2.2.7  The decision rule for this test is to compare W as calculated
in 2.6 to a critical value found in Table 6, Appendix B.  If the computed
W is less than the critical value, conclude that the data are not
normally distributed.  For the data in this example, the critical value
at a significance level of 0.01 and n = 18 observations is 0.858.  Since
W = 0.942 is greater than the critical value, conclude that the data are
normally distributed.

14.2.3  Test for Homogeneity of Variance

14.2.3.1  The test used to examine whether the variation in the percent
of unfertilized eggs is the same across all  copper concentrations
including the control, is Bartlett's Test (Snedecor and Cochran, 1980).  •
The test statistic is as follows:                                         ;
B =
    p
[ (  I
   1=1
                            -„
                         In S2 - i Vi  in
                                1=7
                                    258

-------
     Where V-j  =   degrees of freedom for each copper concen
                  tration and control,  V-j = (n-j  - 1)

           p   =   number of levels of copper concentration
                  including th-e control
           S2 =
1 = 1
           C   =  1  + (  3(p-l))-l  [  I  1/Vi  -  (  I ViH  ]
                                 1=1         1=1

           In  =  loge

           i   =1,2,  ...,  p where p  is the  number of concentrations
                             including the control
           ni  =  the number  of replicates  for  concentration  i.

 14.2.3.2   For the data  in  this  example,  (See Table 4}  all  copper
 concentrations  including the control  have the same number  of  replicates
 (ni  =  3 for all  i).   Thus, Vj = 2 for all  i,

 14.2.3.3   Bartlett's  statistic  is, therefore:

                   ' '   :"':/       P  -""    .                     ;   ;
       B  = [(I2)ln(0.007) - 2  I  ln(Sj)2]/1.194           :   T
                              1=1  ,.

          = [12(-4.962)  -  2(-31.604}]/1.194

          - 3.664M.194

          =3.069

 14.2.3.4   B is  approximately distributed as  chi-square with p-1  degrees
 of  freedom, when  the  variances  are in fact the same.  Therefore, the
 appropriate critical  value for  this  test, at a significance level  of 0.01
;with 5 degrees  of freedom, is 15.09.   Since  B = 3,069  is less than the
 critical  value  of 15.09, conclude that the variances are not  different.
                                    259

-------
 14,2.4.   Dunnett's  Procedure          .•   .

 14.2.4.1  Calculations                         :;

 To obtain an estimate  of  the  pooled  variance  for  the  Dunnett's  Procedure
 construct an ANOVA  table  as described in  Table  8.

                           TABLE  8.  ANOVA  TABLE
Source df . Sum of Squares
(SS)
Between p - 1 . ,,. SSB
Within N - p SSW
Mean Square (MS)
(SS/df)
2
SB = SSB/(P-D
2
SW = SSW/(N-p)
Total N - 1 SST
Where:
p  = number of copper concentrations including the control
N  = total number of observations n-j 4- n;? ••« +np
n-j = number of observations in concentration 1
           SSB = I T-j2/ni .
           SST = i   I YlM2 . G2/N
                1-1  j=l  J

           SSW = SST - SSB
                              Between Sum of Squares



                              Total Sum of Squares


                              Within Sum of Squares
                                                     p
   = the grand total  of all  sample observations, G = I
            G

            T-J  = the total  of the replicate Measurements for
                 concentration
            -JJ  = the  jth  observation for  concentration "i"  (represents
                 the  percent of unfertilized eggs for copper
                 concentration  i  in  test  chamber j)
                                    260

-------
14.2.4.2  For the data in this  example:


    N  =18
    T] = Yn  + Yi2 + Yi3 = 1.255
    T2 = Ygl  + Y22 + Y23 = 1.653
    TS = YSI  + Y32 + Y33 = 1.677
    T4 = Y4i  + Y42 + Y43 = 2.051 .
    T5 = Y5i  + Y52 + Y53 = 2.669
    T6 = Y61  + Y62 + Y63 = 3.244

    G  = 1] + T2 + T3 + T4 =  12.549

          P   o
    SSB = I Tj2/ni - G2/N


        =  1  (28.973) - (12.549)2  = 0.909
                            1
          D   n-i
    SST =
        = 9.740 - (12.549)2  = 0.991
                      18

    SSW = SST - SSB = 0.991 - 0.909 = 0

    SB2 = SSB/p-1 = 0.909/6-1 = 0.182

    SW2 = SSW/N-p = 0.082/18-6 = 0.007
14.2.4.3  Summarize these calculations in the ANOVA table (Table 9}
           Table 9.  ANOVA TABLE FOR DUNNETT'S PROCEDURE EXAMPLE
Source
Between
Within
df
5
12
Sum of Squares
(SS)
0.909
0.082
Mean Square(MS)
CSS/df)
0.182
0.007
    Total
17
0.991
                                    261

-------
  14.2.4.4  To perform the individual comparisons, calculate the t statistic
  for .each concentration, and control combination as follows:
                                  w
                                        (1/ni) + (1/m)
  Where Yi
        n
           - mean percent unfertilized eggs for copper concentration i
           = mean percent unfertilized eggs for the control
           = square root of within mean sqaure
           = number of replicates for control
           = number of replicates for concentration i.

 Since we are looking for an increased response over control  in  percent
 unfertilized eggs, the control  mean is subtracted  from the mean at  a
 concentration.

t!4.2.4.5  Table  10 includes the calculated t  values for each
.concentration and control  combination.   In this example, comparing the
     ug/L concentration with the control the calculation is as follows*
                              ( 0.551 - 0.418  )


                         [ 0.084^  (1/3) +  (1/3) ]
                                                 =  1.956
                        TABLE  10.   CALCULATED T-VALUES
            Copper Concentration (ug/L)
I 2.5
1 5.0
| 10.0 ;:
1 20.0
, 40.0
2
3
4
5
6
1.956
2.074
3.912
6.941
9.750
(14.2.4.6  Since the purpose of this test is to detect a significant
nncrease in percent unfertilized eggs,  a (one-sided)  test  is
-appropriate.  The critical  value for this one-sided  test is found in
liable 5, Appendix C.   For an overall alpha level  of  0.05,  12 degrees of
|freedom for error and five  concentrations (excluding  the control)  the
^critical value is 2.50.   The mean percent unfertilized eggs for
                                     262

-------
I concentration "1" is considered significantly greater than the mean
.percent unfertilized eggs for the control if tj is greater than the
'critical value.  Therefore, the 10.0, 20.0 and 40.0 ug/L concentrations
 have a significantly higher mean percent of unfertilized eggs than the
{control.  Hence the NOEC is 5.0 ug/L and the LOEC is 10.0 ug/L.

 14.2.4.7  To quantify the sensitivity of the test, the minimum
 significant difference (MSD) that can be statistically detected may be
 calculated.
MSD = d SWV"
                                        + U/n)
[Where  d  = the critical value for the Dunnett's procedure
        Sy = the square root of the within mean square
        n  = the common number of replicates at each concentration
             (this assumes equal replication at each concentration
        n-j = the number of replicates in the control.

 14.2.4.8  In this example,

                    MSD = 2.50 (0.084)  V (1/3)  + (1/3)
                        = 2.50 (0.084H0.817)
                        = 0.172

 14.2.4.9  The MSD (0.172) is in transformed units.   To  determine  the  MSD
 in terms of percent survival, carry out the following conversion.
         1.  Subtract  the
         from the transformed control  mean.

        0.418 -  0.172  =  0.590
         2.  Obtain  the untransformed  values  for  the  control mean  and  the
            difference calculated  in  4.10.1.

                          [Sine (0.418) ]2 = 0.165
                   ,:.      [Sine (0.590) ]2 = 0.310

         3.  The  untransformed MSD  (MSDU)  is  determined by subtracting
            the  untransformed values  from  4.10.2.

                        MSDu =  0.310 - 0.165 = 0.145

J4.2.4.10   Therefore,  for this  set of  data, the minimum difference in
jmean percent of unfertilized eggs between the control and any copper
^concentration that can be detected as  statistically significant  is 0.145.
 14.2.4.11  This represents a
'control.
              increase  in unfertilized eggs from the
                                    263

-------
 (4.2.5   Probit  Analysis
 j
J4.2.5.1   The data  used  for  the  probit  analysis  is  summarized  in
gable  11.   To perform the  Probit Analysis,  run the  EPA  Probit  Analysis
Program.   An  example  of  the  program output  is provided  in  Table 12  and
   3. 3.

 [4.2.5.2   For this  example,  the  chi-square  test  for  heterogeneity was not
 ;ignificant.  Thus  Probit  Analysis  appears  to be appropriate for-this set
 Jf  data.
                     TABLE  11.   DATA  FOR  PROBIT ANALYSIS
                          Control
             Copper Concentration (ug/L)

          2.5    5.0    10,0   20.0   40.0
 lumber  Unfertilized
 lumber  Counted
 50
300
 83
300
 85
300
120
300
181
300
232
300
                                      264

-------
FABLE 12. OUTPUT FROM EPA PROBIT ANALYSIS PROGRAM, VERSION  1.3,
          USED FOR CALCULATING EC VALUES
  Probit  Analysis of Sea Urchin Fertilization  Data

                                    Observed      Adjusted       Predicted
               Number     Number   Proportion     Proportion      Proportion
      Cone.    Exposed     Resp.    Responding     Responding      Responding
Control
2 5000
:. 5 00 CO
1C . 0000
20 0000
40 0000
300
300
IOC
3C£
300
300
50
83
85
1 20
181
232
0 . 1667
0.2767
0.2833
0. 4000
0 . 6033
0 . 7733
                                                    0  0000
                                                    0  1085
                                                    0  1167
                                                    0  2605
                                                      5111
                                                      7206
     0 . 1886
     0.0495
     0. 1362
     0.2926
     0- 5025
     0.7117
  Chi  -  Square Heterogeneity
                                  5-238
Hu
Sigma =
Para me t er
Ir-tercepi
Elope ., ,
1 . 297567
0 . 545280
Estimate
2 . 62C368
1.813919


Std Err .
0 . 2346! 4 (
0 1 fi 2 1 2 7 (


9 5% Con f i dence
7 16C525 , 3 .
1 . 47 695C , 2 ,


Limits
080210)
190888)
  Spontaneous
  Response Rate
                  0.188637
                              0.020413
                                                0 . 148627
0.228646)
        Estimated EC Values and Confidence  Limits
  Point
                     Cone .
                                         Lower        Upper
                                       95% Confidence  Limits
EC 1
EC 5
! EC10
t EC15
EC50
i EC85
| EC90
EC95
EC99
00
00 • .''
00
00
00
00
00
00
00
1
2
3
5
19
72
99
156
368
0693
5156
9695
4006
8412
8947
1742
4945
1740
0
1
.:. 2
3
16
57
75
1 1 1
230
4955
4235
4927
6315
7757
8561
5491
6525
6104
1
3
5
7
23
IOC
145
254
729
8C71
7241
4894
1457
1 322
3032
6703
4105
2836

                                     265

-------
  Probi t
    10-t
       EC01
                   EC10    EC25    EC50
Figure 3. Plot of adjusted probits and predicted regression line
                                266

-------
15. PRECISION AND ACCURACY

15.1 PRECISION

15.1.1  Single laboratory precision data for the reference toxicants, copper
(CU) and sodium dodecyl sulfate (SDS) tested in FORTY FATHOMS* artificial
seawater and natural seawater are provided in Tables 13 and 14.  The results
were similar in both types of seawater.  The precision of the SDS data
(30.6% and 35.7%), expressed as a percent coefficient of variation, was
somewhat better than the CU data (44.5% and 48.0%).

15.1.2  No data are available on the multi-laboratory precision of the test.

15.2 ACCURACY

15.2.1 The accuracy of toxicity tests cannot be determined.
                                     267

-------
[TABLE  13.  SINGLE  LABORATORY  PRECISION  OF THE SEA URCHIN  (ARBACIA PUNCTULATA)
           FERTILIZATION  TEST PERFORMED IN FORTY FATHOMS* ARTIFICIAL
           SEAWATER,  USING GAMETES FROM ADULTS MAINTAINED IN FORTY
           FATHOMS^ ARTIFICIAL SEAWATER, OR OBTAINED DIRECTLY FROM
           NATURAL SOURCES, AND COPPER  (CU) AND SODIUM DODECYL SULFATE
           (SDS) AS REFERENCE TOXICANTS^,2,3,4,5
m
m
m
!
m

m
1
i
i
SI
I
P.,
|
2
3
4
5

EC50
(ug/L)
19.1
42.2
27,3
57.8
61.5



14
27
20
47
49

CI
(ug/L)
.9-24.6
.4-56.6
.4-35.3
.2-69.3
.5-77.1
CU
WOEC
(ug/L)
5.0
12.5
< 6.2
6.2
12.5

LOEC
(ug/L)
10.0
25.0
6.2
12.0
25.0

EC50
{mg/L}
1.2
1.9
3.3
2.9
3.0



1
1
2
1
1

CI
(mg/L)
.6-2.4
.4-2.5
.2-4.5
.8-4.0
.9-4.1
SDS
NOEC
(mg/L)
< 0.9
0.9
1.8
0.9
1.8

LOEC
(mg/L)
0,9
1.8
3.6
1.8
3.6
MEAN   41.6 + 44.5%
2.46 + 35.7%
       performed by Dennis McMullen, Technical Applications Inc.,
 Newtown Facility, Environmental Monitoring and Support Laboratory -
 Cincinnati.

2A11 tests were performed using Forty Fathoms^ synthetic seawater.
 Copper test solutions were prepared with copper sulfate.  Copper
 concentrations in Test 1 were: 2.5, 5.0, 10.0, 20.0, and 40.0 ug/L.
 Copper concentrations in Tests 2-5 were: 6.25, 12.5, 25.0,
 50.Os and 100.0 ug/L.

3SDS concentrations were: 0.9, 1.8, 3.6, 7.2, and 14.4 mg/L.
^Adults collected in the field.

5For a discussion of the precision of data from chronic toxicity
 tests see Section 4, Quality Assurance.
                                      268

-------
 TABLE  14
SINGLE LABORATORY PRECISION OF THE SEA URCHIN (ARBACIA PUNCTULATA)
FERTILIZATION TEST PERFORMED IN NATURAL SEAWATER, USING GAMETES
FROM ADULTS MAINTAINED IN NATURAL SEAWATER AND COPPER (CU) AND
SODIUM DODECYL SULFATE (SDS) AS REFERENCE
TOXICANTSl,2,3,4,5
i 	 	 — 	
I x cu
1
jTest
i

1 ^
i *
i
s
EC50 CI
(ug/L) (ug/L)
17.8 17.
45.7 42.
44,6 41.
25,4 23.
16.0 15.
0-18.8
9-48.6
9-47.3
1-28.0
1-17.1
NOEC
(ug/L)
12.2
12.2
24,4
< 6.1
6.1
LOEC
(ug/L)
24.4
24.4
48.7
6.1
12.2
SDS
EC50 CI
(mg/L) (mg/L)
2.6
4.6
2.7
2.4
2.6
2.5-2.6
4.4-4.8
2.7-2.8
2.3-2.5
2.5-2.6
NOEC
(mg/L)
1.8
1.8
1.8
0.9
1.8
LOEC
(mg/L)
3.6
3.6
3.6
1.8
3.6
 Mean   29.9 + 48.
                                   2.98  +  30.
 ]Tests performed by Ray Walsh and Wendy Greene,  Environmental  Research
^  Laboratory,  U.  S.  Environmental  Protection Agency,  Narragansett,  Rhode Island.
j2Copper concentrations were:  6.1, 12.2,  24.4,  48.7,  and  97.4  ug/L.
|3SDS  concentrations were:  0.9,  1.8,  3.6,  7.3,  and  14.5 mg/L.
j4Adults collected in the field.
|5For  a discussion of the precision of  data  from  chronic  toxicity
  tests see,Section  4,  Quality Assurance.
                                       269

-------
    Figure 4.  Data sheet for (1)  fertilization  test using  Arbacia  punctulata.
 •£ST DATE:




 SAMPLE:
 lOMPLEX EFFLUENT SAMPLE:
    COLLECTION DATE:
    SALINITY/ADJUSTMENT:
    PH/ADJUSTMENT REQUIRED:
    PHYSICAL CHARACTERISTICS:



    STORAGE:	



    COMMENTS:
.SINGLE  COMPOUND:
     SOLVENT  (CONC):
    TEST  CONCENTRATIONS:




    DILUTION  WATER: 	



    CONTROL WATER:
     TEST  TEMPERATURE:



     TEST  SALINITY: _



     COMMENTS:
                                        270

-------
  Figure 5.   Data sheet (2)  for  fertilization  test using  Arbacia  punctulata

[TEST DATE:	'

ISAMPLE:
fSPERM DILUTIONS:
     HEMACYTOMETER COUNT,  E:
x TO* = SPM SOLUTION E =
     SPERM CONCENTRATIONS:   SOLUTION  E  x  40  =  SOLUTION A  =
                            SOLUTION  E  x  20  =  SOLUTION B  =
                            SOLUTION  E  x   5  =  SOLUTION C  =
     SOLUTION  SELECTED  FOR  TEST  (

     DILUTION:   SPM/(5  x  10?)  = _
                 [(DF)  x  5)  -  5  =

     FINAL  SPERM COUNTS =
 = 5 x 10? SPM}:

       DF
              + SW,  mL
                                 SPM
                                 SPM
                                 SPM
:EGG  DILUTIONS:

                                              INITIAL EGG COUNT =
     ORIGINAL  EGG  STOCK  CONCENTRATION  =   10X  (INITIAL
                                              EGG COUNT)
     VOLUME  OF  SW  TO ADD TO DILUTE  EGG STOCK  TO 20GO/nt:
                                              (EGG COUNT) - 200 =
                            CONTROL WATER TO ADD EGG STODK, mL =
                                               FINAL EGG COUNT =

ITEST  TIMES:
      SPERM  COLLECTED:

      EGGS COLLECTED:

      SPERM  ADDED: 	

      EGGS ADDED:
      FIXATIVE ADDED:

      SAMPLES READ:
                                     271

-------
   Figure 6. Data sheet (3) for fertilization test using Arbacia punctulata.




 DATE  TESTED:
I            "	™" *" -" «'• • I I  I    ••••! i.n«il H II I  I       || 	• II    I    .•.., •i,||       ,	      -  •
I



'SAMPLE:	;	.	;	








               TOTAL AND  UNFERTILIZED  EGG COUNT  AT END  OF  TEST:
FEFFLUENT

      (%)
                   REPLICATE  VIAL
I
             TOTAL-UNFERT
           TOTAL-UNFERT
TOTAL-UNFERT
TOTAL-UNFERT
STATISTICAL ANALYSIS:



    ANALYSIS OF VARIANCE:



         CONTROL:
         DIFFERENT FROM CONTROL  (P
    COMMENTS:
                                       272

-------
                                     SECTION 16

                                   TEST METHOD^.2

                  ALGAL (CHAMPIA PARVULA) SEXUAL  REPRODUCTION TEST
                                    METHOD  1009
 |.   SCOPE  AND APPLICATION

  ..1   This  method measures the  effects  of  toxic  substances  in  effluents  and
  -eceiving  water on  the sexual  reproduction  of the  marine macroalga,  Champia
 J.arvula.   The method  consists  of  exposing male  and female  plants  to  test
 ^Substances for two  days,  followed by a  5-to 7-day  recovery period in control
 jmediunj, during which  the  cystocarps mature.

                                                               substance
 ;1.3  Single or multiple excursions in toxicity may not be detected using 24-h
 =compositetsamples.  Also, because of the long sample collection period
 (involved in composite sampling, highly volatile and highly deqradable
 toxicants in the source may not be detected in the test.

 1.4  This method should be restricted to use by, or under the supervision of,
 professionals experienced in aquatic toxicity testing and algal culturing.

 2.  SUMMARY OF METHOD
 |                   _                                      .  ..,

 |2.1  Sexually mature male and female branches  of Champia paryula are exposed
 nn astatic system for two days to different concentrations of effluent,  or to
receiving water,followed by a 5-  to 7-day recovery period in control medium.
line recovery period allows time for the  development of  cystocarps resulting
pom fertilization during the exposure period.   The test  results are reported
^s the  concentration  of  the test substance  which  causes a statistically
 ;ignificant  reduction in the number  of cystocarps  formed  compared  to control
 )rganisrns •

     DEFINITIONS

    (Reserved for addition  of  terms at a  later date).

     INTERFERENCES

 -.1  Toxic substances may  be  introduced by  contaminants in dilution water,
 jlassware, sample hardware, and  testing equipment  (see Section 5, Facilities
 md Equipment].

 |The format used for this  method was taken from Kopp, 1983.
 -This method was adapted from Thursby and Steele,  1987,  Environmental
 esearch Laboratory, U. S. Environmental  Protection Agency, Narragansett,
 (node Island.
                   • >          ...     273

-------
  |5.    SAFETY

  5.1   See  Section  3,  Safety and Health.

        APPARATUS AND EQUIPMENT

  j6.1   Facilities for  holding and acclimating test organisms
                                                                ~t.r (DII or
 j6.6  Air pump  -- for  supplying air.

 |6.7  Air lines, and air stones - for aerating cultures.      v

 6.8  Balance - Analytical, capable of accurately weighing to 0.0001 g

 6.9  Reference weights, Class S - for checking performance of balance
                                                                  ,
 parameter,  a portable, field-grade instrument is  acceptable
j6.11   Dissecting (stereoscope)  microscope  -  for  counting  cystocarps.
j6.12   Compound  microscope  -  for  examining  the  condition of plants.
,6.13   Count  register,  2-place -  for recording  cystocarp counts.  ., ,,
                            incubatin9 exposure
                                                                          the
                                     274

-------
  .15  Drying oven -- to dry glassware.
|5.16  Filtering apparatus -- for use with membrane filters (47mm).
  .17  Refractometer -- for determining salinity.
  i.T8  Thermometers, glass or electronic, laboratory grade -- for measuring
  later temperatures.
  ..19  Thermometers, bulb-thermograph or electronic-chart type — for
  Continuously recording temperature.
lf.20  Thermometer, National Bureau of Standards Certified (see USEPA METHOD
  [70.1, USEPA, 1979) -- to calibrate laboratory thermometers.
|.21  Beakers -- Class A, borosilicate glass or non-toxic plasticware, 1000
  iL for making test solutions.
  .22  Erlenmeyer flasks, 250 ml, or 200 ml disposable polystyrene cups, with
  :overs -- for use as exposure chambers.
  ..23  Bottles -- borosilicate glass or disposable polystyrene cups (200-400
|nL) for use as recovery vessels.
  i.24  Wash bottles -- for deionized water, for rinsing small glassware and
finstrurnent electrodes and probes.
  .25  Volumetric flasks and graduated cylinders -- Class A,  borosilicate
Iglass or non-toxic plastic labware, 10-1000 ml for making test solutions.
  ..26  Micropipetters, digital, 200 and 1000 uL -- to make dilutions.
JU.27  Pipets, volumetric — Class A, 1-100 ml.
 I
 |.28  Pipetter, automatic -- adjustable,  1-100 ml.
  .29  Pipets, serological -- 1-10 mi_s graduated.
  .30  Pipet bulbs and fillers -- PROPIPETR, or equivalent.
  ;.31  Forceps, fine-point, stainless steel -- for cutting and handling
 jranch tips.
     REAGENTS AND CONSUMABLE MATERIALS
  f.l  Mature Champla parvula plants -- see Paragraph 7.14.
  .2  Sample containers -- for sample shipment and storage (see Section 8,
 Iffluent and Receiving Water Sampling and Sample Handling).
                                       275

-------
 J7.3  Petri dishes, polystyrene  --  to  hold  plants  for  cystocarp  counts  and  to
 ;cut branch tips.  Other  suitable containers may be  used.
 I
 |7.4  Disposable tips for micropipetters.

  .5  Aluminum foil, foam stoppers, or other closures  -- to cover  culture and
Itest flasks.
  .6  Tape, colored — for  labelling test  chambers.

  .7  Markers, water-proof  -- for marking  containers, etc.

  .8  Data sheets  (one set  per test) -- for data recording.

  .9  Buffers, pH  4, 7, and 10 (or as per  instructions of  instrument
  lanufacturer) for standards and calibration check  (see USEPA Method  150.1,
 |JSEPA, 1979).
 i
 [7.10  Laboratory  quality assurance samples and standards, for the above
 [methods.

 p.11  Reference toxicant solutions (see Section 4, Quality Assurance).

 17.12  Reagent water — defined as distilled or deionized  water that  does  not
 fcontain substances which are toxic to the test organisms  (see paragraph 6.5
 above).

 •7.13  Effluent, surface water, and dilution water  -- see  Section 7,  Dilution
 jjWater, and Section 8, Effluent and Surface Water Sampling and Sample
 Dandling.

 |7.13.1  Saline test and dilution water -- The use  of natural seawater is
 "recommended for this test.  A recipe for  the nutrients that roust be  added to
  ;he natural sea water is given in Table 1.  The salinity  of the test water
  lust be 30 °/oo,  and vary  no more than ±  2 °/oo among the replicates.

 |7.13.2  The overwhelming majority of industrial and sewage treatment
 feffluents entering marine  and estuarine systems contain  little or no
 ^measurable salts.  Therefore, exposure of Champia  parvula to effluents will
 'usually require adjustments in the salinity of the test solutions.   Tests
  lust be performed with a minimum of 50% natural seawater  at each toxicant
  :oncentration.  It is important to maintain a constant salinity across all
 ^treatments.  The  salinity  of the effluent can be adjusted by adding  brine
  >repared from natural seawater (100 °/oo), concentrated  (triple strength)
 |salt solution (GP-2 described in Table 2), or dry  GP-2 salts (Table  2), to
 the effluent to provide a  salinity of 30  °/oo.  Control solutions should
    prepared with  the same  percentage of natural seawater  and at the  same
 salinity (using deionized  water adjusted with dry  salts,  or brine) as used
 |for the effluent  dilutions.

I/.13.3  Artificial seawater -- The preparation of  artificial seawater (GP-2)
  s described in Table 2.

                                       276

-------
  7.14   CHAMPIA  PARVULA  CULTURES

  7.14.1  Mature  plants (Figure  1) are available from the Environmental
  Research Laboratory, U. S. Environmental Protection Agency, South Ferry
  Road, Narragansett, Rhode  Island, 02882  (401-782-3000).
  tetraaporongio
                                                             spermatia
                                                            fertilization
        TETRASPOROPHYTE
                                                           —cystocarp
  'igure 1. Life history of Champia parvula.   Upper left:  Size and degree of
 Branching in female branch tips used for toxicity tests.   From Thursby and
|5teeie, 1987.

                                      277

-------
 The  adult  plant  body  (thallus}  is  hollow,  septate, and highly branched.  New
 cultures can  be  propagated  asexually from  excised branches, making  it possible
 to maintain clonal  material  indefinitely.

 7.14.1.1   Stock  cultures of  both  male and female plants are maintained  in
 separate,  aerated,  1-L  Erlenmeyer  flasks,  or equivalent, containing 800  ml of
 culture medium.   Several cultures  of both  males and females are maintained
 simultaneously to keep  a constant  supply of plant material available.  New
 cultures should  be  started  weekly, so that plants are available in different
 stages of  development  (i.e.,  with  different amounts of tissue per flasks).  The
^total number  of  cultures maintained will depend on the expected frequency of
^testing.

p.14.1.2   Prior  to  use  in toxicity tests,  stock cultures should be examined to
^determine  their  condition.  Females can be checked by examining a few branch
[rips  under a  compound microscope {100 X or greater).  Several trichogynes
 (reproductive hairs to  which  the spermatia attach) should be easily seen near
fthe  apex {Figure  2).
                                        sterile heirs
                                               — trichogynes
                         1 mm
Figure 2.  Apex of branch of female plants showing sterile hairs and
reproductive hairs (trichogynes).  Sterile hairs are wider and generally much
longer than trichogynes, and appear hollow except at the tip.  Both types of
hairs occur on the entire circumference of the thallus, but are seen easiest at
the "edges."  Receptive trichogynes occur only near the branch tips.  From
Thursby and Steele, 1987.

                                       278

-------
 7.14.1.3  Male plants should be visibly producing  spermatic.  This can be
 checked by placing some male tissue in a petri  dish,  holding  it against a
 dark background and looking for the presence of spermatial sori.  Mature
 sori can also be easily identified by looking along the edge  of the thallus
 under a compound microscope (Figures 3 and 4).
                    1 cm
                                     ^•spermotial
                                                sorus
 Figure 3.  A portion of the male thallus showing  spermatial sori.  The
•sorus areas are generally slightly thicker and somewhat  lighter in color.
From Thursby and Steele, 1987.
                             •cuticle
                                         thailus surface
  Figure 4.
A magnified portion of a  spermatial sorus.  Note the rows of
cells that protrude from  the  thallus surface.  From Thursby and
Steele, 1987.
                                      279

-------
|f                                                                          'SlfflH
17.14.1.4  A final, quick way to determine  the relative  "health"  of  the maff|
'.stock culture  is to place a portion of a female  plant  into some  of  the water
 from the male  culture for a few seconds.   Under  a  compound microscope
 numerous spermatia should be seen attached to both the  sterile hairs and  the
 trichogynes (Figure 5).
  'igure 5.  Apex of a branch on a mature female plant that was exposed to
[spermatia from a male plant.  The sterile hairs and trichogynes are covered
[with spermatia.  Note that few or no spermatia attached to the older hairs
 [(those more than 1 mm from the apex).  From Thursby and Steele, 1987.
  .14.2  Culture medium prepared from natural seawater is preferred
 l(Table 1).  However, as much as 50% of the natural seawater may be replaced
    the artificial seawater (GP-2) described in Table 2.

  .14.2.1  The seawater is autoclaved for 20 min at 15 psi. The culture
  Tasks are capped with aluminum foil and autoclaved dry, for 10 min.
 ;ulture medium is made up by dispensing seawater into sterile flasks and
 idding the appropriate nutrients from a sterile stock solution.

 '.14.2.2  Alternately, 1-L flasks containing seawater can be autoclaved.
 iterilization is used to prevent microalgal contamination, and not to keep
 :ultures bacteria free.

     SAMPLE COLLECTION, PRESERVATION AND HANDLING

 U   See Section 8,  Effluent and Receiving water Sampling and Sample Handling

     CALIBRATION AND  STANDARDIZATION

[9.1   See Section 4,  Quality Assurance.                               :
I                                                                ''••',.
110.   QUALITY CONTROL

|10.1   See Section 4, Quality Assurance.

                                       280

-------
 TABLE 1. NUTRIENTS TO BE ADDED TO NATURAL SEAWATER AND TO ARTIFICIAL
        '  SEAWATER (GP-2) DESCRIBED-IN TABLE 2.  THE CONCENTRATED NUTRIENT
          STOCK SOLUTION IS AUTOCLAVED  FOR 15 MIN (VITAMINS ARE
          AUTOCLAVED SEPARATELY FOR 2 MIN AND ADDED AFTER THE NUTRIENT
          STOCK SOLUTION IS AUTOCLAVED).  THE PH OF THE SOLUTION IS
          ADJUSTED TO APPROXIMATELY PH 2 BEFORE AUTOCLAVING TO MINIMIZE
          THE POSSIBILITY OF PRECIPITATION.
                                Amount of Reagent Per Liter of Concentrated
                                       Nutrient Stock Solution
                                Stock Solution For
                                 Culture Medium
                     Stock Solution For
                        Test Medium
 Nutrient Stock Solution^

 Sodium Nitrate
 (NaN03)

 Sodium Phosphate
 (NaH2P04- H20

 Na2EDTA - 2 H20

 Sodium Citrate
6.35 g


0.64 g

 133 mg


  51 mg
 1.58 g


 0.16 g




12,8 mg
 Iron2

 Vitamins3
 Udd  10  ml  of  appropriate  nutrient  stock  solution  per  liter  of  culture  or
  test  medium.

,2A  stock solution  of iron  is  made that  contains  1  mg iron/mL.   Ferrous  or
 ferric chloride  can  be  used.   Add 9.75  mL  of  the iron  stock  solution  to each
 liter  of culture medium and 2.4  mL  to each  liter of test medium.

|,3A  vitamin  stock solution  is  made by dissolving  4.88. g  thiamine HC1,
  2,5 mg  biotin,  and  2.5 mg B12 in 500 mL  deionized water.  Adjust  to
  approximately pH  4  before autoclaving  2 min.  It  is convenient to
  subdivide  the vitamin  stock  into 10 mL volumes  in test tubes prior to
  autoclaving.  Add  10 ml of the vitamin  stock  solution to each liter of  culture
  medium  and 2.5  mL to each liter of test medium.
                                       281

-------
 TABLE  2.
            SALTS USED IN THE PREPARATION OF  GP-2 ARTIFICIAL SEAWATER
                        S                             ALSEAWATER
           Compound
                                                Concentration (g/L)

                                                       21.03
                                                        3.52
                                                        0.6]
                                                        0.088
                                                        0.034
                                                        9.50
                                                        1.32
                                                        0.02
                                                        0.17

iThe original  formulation  calls for  autoclaving anhydrous and hydrated
 salts  separately to  avoid  precipitation.  However, if the sodium
 SKV8  ?"toc1a^d  separately  (dry), all of the other salts can be
 is  nSJ  cHt5S? Kr'a«]nC%n; n*Utn'ent? are ddded unt11 needed> autoclaving
 IL  \•*     ?    °r  effluent testing.  To minimize microalgal contamination
 the artificial  seawater should be autoclaved when used for stock culture
                      f°r  * ^ ]°          "           and        ™
I NaCI
I-
1 Na2S04
1 KC1
1 KBr
I Na2B407
HgCl2 •
f CaCl2 -
SrCl2 •
NaHC03
— • • — — • — -



• 10 H20
6 H20
2 H20
6 H20

,2Prepare in 10-L to 20-L  batches.
J3A stock solution of 68 mg/ml  sodium bicarbonate is prepared by autoclaving
  lar^Ht^nf0^'  3nd  ohcn  ,disS°1Jn'ng  n  1n sterile ^Ionized water.  For
  each  liter of  GP-2,  use  2.5 ml of this stock solution.
prom  Spotte, et al.,  1984.
Affluent  salinity adjustment  to 30 o/00 can be made by adding the
5  appropriate amount  of dry salts from this formulation, by using a
  ™npirstre??th  br1ne PrePared from this formulation,  or by using a
  100 o/oo  salinity brine prepared from natural seawater.
'Nutrients  listed  in Table 1 should be added to the artificial  seawater in
  the same concentration described for natural seawater.
                                      282

-------
   11.  TEST PROCEDURES
   11.1  TEST SOLUTIONS
   11.1.1  Surface Waters
 collected.
   11.1.2  Effluents
                            tOXlC1'ty 1S  determined  witb  samples  used  directly as
          t«t         °f  th!  re(lul>eillent t° use a minimum of 50% natural seawater
          test  solutions,  the concentration of effluent used in this test if
    " ^                                       '       "                 eved
                 nf th  +  3! thn eff]uent test concentrations should be based on
  0.5, is conmonlv u ed  Tdi1u?L f^ d11Jt1on faCt°rS' Wox1n,ately 0.3 or
  betwPPn Rni ann i* fJii   dilution factor of approximately 0.3 allows testing
  II  9*  ,nH ni«\  e;;|uent using only five effluent concentrations (50%,  17«,
  6%, 2%, and 0.7%).  This series of dilutions minimizes the level  of effort
  tP but it is not critical for the recovery
                "? U1"?  seawater ^all'ty can vary among laboratories,  a more
                nt medfUm (e'9' + EDTA) m* result in "«ter  growth  and
 therefore faster cystocarp development) during the recovery  period.
 11.2  PREPARATION OF PLANTS FOR TEST
 11.2.1   Once cultures are  determined to be  usable for  toxicity  testina (have
 trie ogynes, and sori with  spermatia),  plant  cuttings  shoud  be prepared for
jthe  test,  using  a fine-point  forceps,  with  the plants  in a littleP  seawater  in
                                       283

-------
a petri dish.  For female plants, five cuttings, severed 7-to-10 mm from the
ends of the branch, should be prepared for each treatment chamber.  Try to
be consistent in the number of branch tips on each cutting.  For male
plants, one cutting, severed 1.5-to-2 cm from the end of the branch, is
prepared for each test chamber.  Prepare the female cuttings first, to
minimize the chances of contaminating them with water containing spermatia
from the male stock cultures.

11.3.  STAKT OF TEST

11.3.1  The test should begin as soon as possible, preferably within 24 h of
sample collection.  If the persistence of the sample toxicity is not known,
the maximum holding time (time elapsed from the removal of the sample from
the sampling device) should not exceed 36 h.  In no case should the test be,
started more than 72 h after sample collection.  Just prior to testing, thMi
temperature of the sample should be adjusted to that of the test           '";
(23 + 1°C) and maintained at that temperature until portions are added
to the dilution water.

11.3.2  Set up and label four test chambers (minimum of three) per treatment
and controls.

11.3.3  Fill the test chambers with 100 mL of control or treatment water
(28-to-32 o/oo).  For reference toxicant tests, all test chambers can be
filled with control water and the toxicant added with a pipet.  For toxicant
volumes exceeding 1 mL, adjust the amount of dilution water to give a final
volume of 100 mL.

11.3.4  Add five female branches and one male branch to each test chamber.
The toxicant must be present before the male plant is
11.3.5  Place the test chambers under cool-white light (approx.
100 uE/m2/s, or 500 ft-c), with a photoperiod of 16 h light and  8 h
darkness.  Maintain the temperature between 22 and 24°C.   Check  the
temperature by placing a laboratory or recording thermometer in  a flask of
water among the test chambers.  Record the temperature daily.

11.3.6  Gently hand-swirl the chambers twice a day, or shake continuously at
100 rpm on a rotary shaker.

11.3.7  If desired, the media can be changed after 24 h.

11.4  TRANSFER OF PLANTS TO CONTROL WATER AFTER 48 H

11.4.1  Label the recovery vessels.  These vessels can be almost any type of
container or flask containing 100 to 200 mL of seawater and nutrients (see
Tables 1 and 2).  Smaller volumes can be used, but should be checked to make
sure that adequate growth will occur without having to change the medium.
                                     284

-------
11.4.2  With forceps, gently remove the female branches from test chambers
and place into recovery bottles.  Add aeration tubes and foam stoppers.

11.4.3  Place the vessels under cool-white light (at the same irradiance as
the stock cultures) and aerate for the 5- to 7-day recovery period.  If a
shaker is used, do not aerate the solutions (this will enhance the water
motion).

11.5  TERMINATION OF THE TEST  .

11.5.1  At the end of the recovery period, count the number of cystocarps
{Figs. 6, 7, and 8) per female and record the data (Figure 12).  Cystocarps
may be counted by placing females between the inverted halves of a      a
polystyrene petri dish or other suitable containers with a small amount <|f
seawater (to hold the entire plant in one focal plane).  Cystocarps can be
easily counted under a stereomicroscope, and are distinguished from young;
branches, because they possess an apical opening for spore release  (ostiole)
and darkly pigmented spores.

11.5.2  One advantage of this test procedure is that if there  is uncertainty
about the identification of an  immature cystocarp, it  is only  necessary to
aerate the plants a  little  longer in the recovery bottles.  Within 24 to
48 h, the presumed cystocarp will either look more like a mature cystocarp
or a young branch, or will  have changed very little, if at all  (i.e., an
aborted cystocarp).  No new cystocarps will form since the males have been
removed, and the plants will only get larger.  Occasionally, cystocarps will
abort, and these should not be  included in the counts.  Aborted cystocarps
are easily  identified by their  dark pigmentation and,  often, by the
formation of a new branch at the apex.

12.  SUMMARY OF TEST CONDITIONS

12.1   A summary of  test conditions  is  listed  in Table 3.

13.  ACCEPTABILITY OF TEST  RESULTS

13.1  A test  is not  acceptable  if the control mortality exceeds  20%.
(Generally  there  is  no control  mortality.)             ..

13.2  If plants fragment  in the controls or  lower exposure concentrations,
it may  be an  indication  that they are under  stress.

13.3  Control  plants should average  10  or  more  cystocarps.
                                      285

-------
                                          ostiole
                                        -spores
                        1mm
Figure 6. A mature cystocarp.   In the controls and lower effluent
concentrations,  cystocarps often occur in clusters of 10 or  12.  From Thursby
and Steele, 1987.
                                         young branch
                                          cells
                                                immature
                                                 cystocarp
Figure 7. Comparison  of  a very young branch and an immature  cystocarp.  Both
can have sterile tiairs*  Trichogynes might or might not be present on a young
branch, but are never present on an immature cystocarp.  Young  branches are
more pointed at the apex and are made up of larger cells than  immature
cystocarps, and never have ostioles.  From Thursby and  Steele,  1987.
                               1mm
Figure 8.   An aborted  cystocarp.  A new branch will  eventually develop at the
apex.  From Thursby and  Steele, 1987.
                                     OQC

-------
   TABLE  3.  SUMMARY  OF  RECOMMENDED  TEST  CONDITIONS  FOR
             CHAMPIA  PARVULA SEXUAL  REPRODUCTION  TEST
 1.  Test  type:

 2.  Salinity:

 3.  Temperature:

 4.  Photoperiod:

 5.  Light intensity:

 6.  Light source:

 7.  Test  chamber:


 8.  Test  solution  volume

 9.  Dilution water:
10.  Dilution factor:

11.  Number of Dilutions

12.  Number of replicate Chambers
      per treatment:

13.  Number of organisms
      per test chamber:

14.  Test duration:
15. Effects measured:
Static, non-renewal

30 o/oo + 2 o/oo

22 - 24°C

16 h light, 8 h dark

100 uE/m2/s (500 ft-c)

Cool-white fluorescent lights

200 mL polystyrene cups, or
250 mL Erlenmeyer flasks

100 mL

30 °/oo salinity natural seawater,
or a combination of 50% 30 °/oo
salinity natural seawater and 50%
30 o/oo salinity artificial seawater

Approximately 0.3 or 0.5

At least 5 and a control


4 (minimum of 3)

5 female branch tips and
1 male plant

2-day exposure to efflent,
followed by 5- to 7-day recovery
period in control medium for
cystocarp development

Reduction in cystocarp production
compared to controls

-------
14.  DATA ANALYSIS

14.1  GENERAL

14.1.1  Tabulate and summarize the data.
Is listed in Table 4.
A sample set of reproduction data
14.1.2  The end points of the Champia parvula toxicity test are based on the
adverse effects on sexual reproduction.  Statistically significant
differences in the mean number of cystocarps, yielding NOEC and LOEC end
points, are determined in most cases by a hypothesis test such as Dunnett's
Procedure (Dunnett, 1955) or Steel's Many-one Rank Test (Steel, 1959;
Miller, 1981).                                               : ,;

14.1.3  Formal statistical analysis of the data is outlined in Figure 9.
Concentrations that have exhibited no sexual reproduction (less than 5% of
controls) are excluded from the statistical treatment of the test data for
calculation of the NOEC and LOEC by Dunnett's Procedure or Steel's Many-one
Rank Test.  The response used in the statistical tests is the mean number of
cystocarps.

14.1.4  When equal numbers of replicates occur across all concentrations and
the control, the statistical analysis consists of a parametric test,
Dunnett's Procedure, and a non-parametric test, Steel's Many-one Rank Test.
The underlying assumptions of the Dunnett's Procedure, normality and
homogeneity of variance, are formally tested.  The test for normality is the
Shapiro-Wilks Test and Bartlett's Test is used to test for homogeneity of
variance.  Tests for normality and homogeneity of variance are included in
Appendix B.  If either of these tests fail, the non-parametric test, Steel's
Many-one Rank Test, is used to determine the NOEC and LOEC end points.  If
the assumptions of Dunnett's Procedure are met, the end points are
determined by the parametric test.

14.1.5  Additionally, if unequal numbers of replicates occur among the
concentration levels tested there are parametric and non-parametric
alternative analyses.  The parametric analysis is the Bonferroni t-test. The
Wilcoxon Rank Sum Test with the Bonferroni adjustment is the non-parametric
alternative.  For detailed information on the Bonferroni adjustment, se.e
Appendix D.

14.1.6  The statistical tests described here must be used with a knowledge
of the assumptions upon which the tests are contingent.  The assistance of a
statistician is recommended for analysts who are not proficient in
statistics.
                                      288

-------
            TABLE 4.  DATA FROM CHAMPIA PARVULA EFFLUENT TOXICITY TEST,
                      CYSTOCARP COUNTS FOR INDIVIDUAL PLANTS AND MEAN
                      COUNT PER TEST CHAMBER FOR EACH EFFLUENT
                      CONCENTRATION 1
Effluent
Concentration
(%)

Control


0.8%


1.3%


2.2%


3.6%


6.0%


10.0%
i
Replicate
Test
Chamber
A
B
C
A
B
C
A
B
C
A
B
C
A
B
C
A
B
C
A
B
C
Plant
1
19
19
17
10
12
12
10
6
4
1
7
3
Z
3
0
1
1
0
0
1
2
2
20
12
25
16
10
9
0
4
4
2
9
2
1
4
4
0
2
4
0
0
1
3
24
21
18
11
6
9
3
4
2
5
9
2
1
6
3
0
1
3
0
0
0
4
7
11
20
12
9
13
5
8
6
4
4
0
5
4
1
0
0
1
0
0
0
5
18
23
16
11
10
8
4
4
4
0
6
0
0
2
3
0
0
3
_
0
0
Mean
Cystocarp
Count
17.60
17.20
19.20
12.00
9.40
10.20
4.40
5.20
4.00
2.40
7.00
1.40
1.80
3.80
2.20
0.20
0.80
2.20
0.00
0.20
0.60
bata provided by the Environmental Research Laboratory, U. S. Environmental
Protection Agency, Narragansett, Rhode Island.
                                      289,.

-------
14.2  EXAMPLE OF ANALYSIS OF CHAMPIA REPRODUCTION DATA

14.2.1  In this example, the data, mean and standard deviation of the
observations at each concentration including the control are listed in Table
5.  The data are plotted in Figure 10.  As can be seen from the data in the
table, mean reproduction per chamber in the 10% effluent concentration is less
than 5% of the control.  Therefore the 10% effluent concentration is not
included in the subsequent analysis.
               TABLE 5.  CHAMP IA PARVULA SEXUAL REPRODUCTION DATA
                                   Effluent Concentration (%)
Replicate    Control
0.8
1.3    2.2
      3.6
      6.0  ' 10



A
B
C
Mean(Yi)
Si2
i


17
17
19
18
1
1
.60
.20
.20

.12

12
9
10
10
1
2
.00
.40
.20
.53
.77

4.40
5.20
4.00
4.53
0.37
3
2.40
7.00
1.40
3.60
8.92
4
1.80
3.80
2.20
2.60
1.12
5
0.
0.
2.
1.
1.
6
20
80
20
07
05

0.00
0.20
0.60
0.27
0.09
7
14.2.2  Test for Normality

14.2.2.1  The first step of the test for normality is to center the
observations fay subtracting the mean of all the observations within a
concentration from each observation in that concentration.  The centered
observations are summarized in Table 6.
    TABLE 6.  CENTERED OBSERVATIONS FOR SHAPIRO-WILKS EXAMPLE
Replicate    Control
                                 Effluent Concentration^)
0.8
 1.3
2.2
3.6
6.0
A -0
B -0
C 1
.40
.80
,20
1.
-1.
-0.
47
13
33
-0
0
-0
.13
.67
.53
-1
3
-2
.20
.40
.20
-0
1
-0
.80
.20
.40
-0
-0
1
.87
.27
.13

-------
            STATISTICAL ANALYSIS OF CHAMPIA PARVULA SEXUAL
                            REPRODUCTION TEST
                             REPRODUCTION DATA
                            MEAN CYSTOCARP COUNT
                            SHAPIRO-WILKS TEST
                                                NON-NORMAL DISTRIBUTION
                 NORMAL DISTRIBUTION
     HOMOGENEOUS VARIANCE
            NO
                               BARTLETT'S TEST
             HETEROGENEOUS
                VARIANCE
                   EQUAL NUMBER OF
                     REPLICATES?
EQUAL NUMBER OF
  REPLICATES?
                     YES


  YES
T-TEST WITH
BONFERRONI
ADJUSTMENT



DUNNETT'S
TEST

•
STEEL'S MANY-ONE
RANK TEST



WILCOXON RANK SUM
TEST WITH
BONFERRONI ADJUSTMENT


                               ENDPOINT ESTIMATES
                                    NOEC. LOEC
Figure 9.  Flow chart for statistical  analysis of Champla parvula data,

-------
  l/JED
Ziu
                                                                                                              o
                                                                                                              o
                                                                                                              o
                                                                                                              (O
                                                                                                              en

-------
 14.2.2,2  Calculate the denominator, D9 of the test statistic

               "•."•-,--•         n
D -
                                  _
                              i - x)2
     Where Xj = the ith centered observation
           X  = the overall mean of the centered observations
           n  = the total number of centered observations.
            n s 18

            X =  1 (0.01)  - 0.0006
 For this set of data,
                                  D - 28.7201

 14.2.2.3   Order the centered observations from smallest to largest
 Where  xH)  is  the  ith  ordered  observation.   These ordered observations
 are  listed  in  Table  7.
   TABLE  7.   ORDERED  CENTERED  OBSERVATIONS  FOR  SHAPIRO-WILKS  EXAMPLE
i
1
2
3
4
5
6
7
8
9
x(D
-2.20
-1.20
-1.13
-0.87
-0.80
-0.80
-0.53
-0.40
-0.40
i
10
11
12
13
14
15
16
17
18
;;- x(D
-0.33
-0.27
-0.13
0.67
1.13
1.20
1.20
1.47
3.40
14.2.2.4 From Table 4, Appendix B, for the number of observations, n,
obtain the coefficients a], a2, ..., ak where k is approximately
n/2.  For the data in this example, n = 18, k = 9.  The a, values are
listed in Table 8.

-------
 14.2.2.5  Compute the test statistic,  W, as follows:
                    i    k
                W = ~ [ I ai (xfn
the differences x(n-
of data,
                     w =
                                 are listed in Table 8.   For this set

                                  (5-U25)2 = °-921
    TABLE  8.   COEFFICIENTS  AND  DIFFERENCES FOR  SHAPIRO-WILKS  EXAMPLE
1
2
3
4
5
6
7
8
9
0.4886
0.3253
0.2553
0.2027
0.1587
0.1197
0.0837
0.0496
0.0163
5.60
2.67
2.33
2.07
1.93
1.47
0.40
0.13
0.07
XH8)
X
-------
           $2  =
                         i  Sj2)
           In = loge
           1  = 1, 2, ..., p where p is the number of concentrations
                             including the control
           rij = the number of replicates for concentration i.

 14.2.3.2  For the data in this example (See Table 5) all effluent

                                               same number of replicates
 14.2,3.3  Bartlett's statistic is therefore:


        B =  [U2Mn(2.39l7J  - 2  £ ln(S1)2]/K1944
          =   [12(0.8720)  -  2(ln(1.12)+ln(1.77)+...+ln(1.05))]/1.1944

          =   (10.4640  - 4.0809)71.1944

          =   5.34

14.2.3.4  B  is approximately distributed as chi-square with p - 1 deqrees
of freedom, when the  variances are  in fact the same.  Therefore, the
appropriate critical  value for this test, at a significance level of 0.01
with five degrees of  freedom, is 15.09.  Since B = 5.34 is less than the
critical value of 15.09, conclude that the variances are not different

14.2.4  Dunnett's Procedure

14.2.4.1 Calculations

To obtain an estimate of the pooled variance for the Dunnett's Procedure
construct an ANOVA table as described In Table 9.              rroceaure,

-------
                           TABLE 9.  ANOVA TABLE
    Source
    Total
       df
Sum of Squares
     (SS)
      N - 1
     SST
Mean Square(MS)
    (SS/df)
HH
HB
Hra Between

H Within
BK 	

P

N

- 1

- P

SSB

SSW
2
SB =
2
SW =

SSB/(p -

SSW/(N -

1)

P) ;
m 	
Hi Where:
B9H
H
m
p
N
n-i
= number effluent concentrations including the control.
= total number of observations n-| + n2 ... +np.
= number of observations in concentration i.
      P
SSB = Z
                                          Between Sum of Squares
                     n-f
           SST = I   I Yt12 - Q2/N
                  i   • •>   "
                1=1  J=1

           SSW = SST - SSB
                               Total Sum of Squares


                               Within Sum of Squares
            G  = the grand total of all sample observations, G = Z T-f
                                                                i = l
            Tj = the total of the replicate measurements for
                 concentration "i"
           Yjj = the jth observation for concentration "i" (represents
                 the mean (across plants) number of cystocarps for
                 effluent concentration i in test chamber j)

14.2.4.2  For the data in this example:
"1
N
T]
T2
TS
T4
T5
T6
G
= n2 = n3 = r
= 18
= Y]] + Yi2 -i
= Y2] + Y22 i
- Y31 + Y32 H
= Y4i + Y42 H
= YSI + Ys2 H
s Y61 + Ye2 H
= T-j + T2 + 1
14 = ns = ng =
^ Y13
•• Y23
^ Y33
•• Y43
»• Y53
^ Y63
[3 + 1
= 17
= 12
= 4
— 0
= 1
= 0
r^ +
*
.
*
*
•
•
6 +
0 +
4 +
4 +
8 +
2 -*•
"7
17
9
5
7
3
0
=
.2
.4
.2
.0
.8
.8
1"5 + ^6 =
3
+ 19.
+ 10.
+ 4.
+ 1.
+ 2.
+ 2.
121.0

2
2
0
4
2
2


= 54
= 31
= 13
= 10
— 7
= 3



.6
.6
.8
.8
.2

                                    296

-------
   SSB = £ Tj2/ni - G2/N
        i = l

       = J_  (4287.24) -  (121.0)2 = 515.59
          3                 18
      P   ni
SST = I   I
     1=1  =l
                     - Q2/N
        =  1457.8  -  (121.0)2  = 644.41
                      18

    SSW =  SST  - SSB  =  644.41 - 615.69 = 28,72

        =  SSB/p-1  =  615.69/6-1 =  123.14

      * =  SSW/N-p =  28.72/18-6 =  2.39

14.2.4.3  Summarize  these  calculations  in  the ANOVA table  (Table  10).


          TABLE  10.  ANOVA TABLE FOR DUNNETT'S PROCEDURE EXAMPLE
Source
Between
Within
Total
df
5
12
17
Sum of Squares
(SS)
615.69
28.72
644.41
Mean Square(MS)
(SS/df)
123.14
2.39

14,2.4.4  To perform the individual comparisons, calculate the t
statistic for each concentration, and control combination as follows:
                          SwvMl/ni) + HAM)


Where Yi  = mean number of cystocarps for effluent concentration i
      YI  = mean number of cystocarps for the control
      $W  = square root of within mean sqaure
      ni  = number of replicates for control
      n-j  = number of replicates for concentration i.

                                     297

-------
14.2,4.5  Table 11 Includes the calculated t values for each
concentration  and control  combination.   In this example,  comparing the
0.8% concentration with the control the  calculation is  as  follows:
                            (   18 - 10.53 )
                  t2 =
              -5.90
                       [  1.55 V (1/3)  f (1/3)   ]
                       TABLE  11.  CALCULATED T-VALUES
           Effluent Concentration^)
0.8
1.3
2.2
3.6
6.0
2
3
4
5
6
                                                        5.90
                                                        0.64
                                                        1.38
                                                        2.17
                                                        3.38
14.2.4.6  Since the purpose of this test is to detect a significant
reduction in cystocarp production,  a (one-sided)  test is appropriate.
The critical value for this one-sided test is found in Table 5,
Appendix E.   For an overall alpha level of 0.05,  12 degrees of freedom
for error and five concentrations (excluding the  control)  the critical
value is 2.50.  Mean cystocarp production for concentration "i"  is
considered significantly less than  mean cystocarp production for the
control  if t^ is greater than the critical value.  Therefore, all
effluent concentrations in this example have significantly lower
cystocarp production than the control.   Hence the NQEC is  Q% and the LOEC
is 0.8%.

14.2.4.7  To quantify the sensitivity of the test, the minimum
significant  difference (MSU) that can be statistically detected  may be
calculated.
                   MSD 7 d. SWv/ (1/ni) +

Where  d  = the critical value for the Dunnett's procedure
       Sy = the square root of the wjthin mean square
       n  = the common number of replicates at each concentration
            (this assumes equal replication at each concentration
       n-j = the number of replicates in the control.
14.2.4.8  In this example,

                   MSD - 2.50
                       = 2.50
                       = 3.16
(1.55)
(1.55
  (1/3)  +
.8165)
                                               (1/3)
                                    298

-------
 14.2.4.9   Therefore,  for  this  set  of  data,  the minimum difference that
 can  be  detected  as  statistically significant  is 3.16 cystocarps.

 14.2.4.10   This  represents a 17.6% reduction  in cystocarp production from
 the  control.

 15.   PRECISION AND  ACCURACY

 15.1  PRECISION

 15.1.1  The single  laboratory  precision data from six tests with copper
 sulfate (CU) and six  tests with sodium dodecyl sulfate (SDS) are listed
;in Table 12.  The NOECs with CU differed by only one concentration
 interval (factor of two), showing  good precision.  The precision of the
 first four tests with SDS was  somewhat obscured by the choice of toxicant
 concentrations, but appeared similar to that of CU in the last two tests.

 15.1.1  The multilaboratory precision of the test has not yet been
 determined.

 15.2  ACCURACY

 15.2.1 The accuracy of toxicity tests cannot be determined.
                                    299

-------
  TABLE 12. SINGLE LABORATORY PRECISION OF THE CHAMP IA PARVULA
            REPRODUCTION TEST PERFORMED IN A 50/50 MIXTURE OF NATURAL
            SEAWATER AND GP-2 ARTIFICIAL SEAWATER, USING GAMETES FROM
            ADULTS CULTURED IN NATURAL SEAWATER.  THE REFERENCE
            TOXICANTS USED WERE COPPER (CU) AND SODIUM DODECYL
            SULFATE (505)1,2,3,4
CU

Test
1
2
3
4
5
6
NOEC
(ug/L)
1.0
1.0
1.0
1.0
0.5
0.-5
LOEC
(ug/L)
2.5
2.5
2.5
2.5
1.0
1.0
SDS
NOEC
(mg/L)
< 0.8
0.48
< 0.48
< 0.48
0.26
0.09

LOEC
(mg/L)
0.8
0.8
0.48
0.48
0.43
0.16
  ^Tests performed by Glen Thursby and Mark Tagliabue,  Environmental
   Research Laboratory, U, S.  Environmental Protection  Agency,
   Narragansett,  Rhode Island.  Tests were conducted at a  temperature
   of 22°C, in 50/50 GP2 and natural  seawater at a salinity of
   300/00.

  2Copper concentrations were: 0.5, 1.0,  2.5, 5.0, 7.5, and
   lO.O.ug/L.

  3SDS concentrations for Test 1  were:   0.8,  1.3,  2.2,  3.6, 6.0,  and
   10.0 mg/L.   SDS concentrations for Tests 2,  3,  and 4 were: 0.48,
   0.8, 1.3, 2.2,  3.6, and 6.0 mg/L.   SDS concentrations for Tests  5
   and 6 were:  0.09, 0.16, 2.26,  0.43,  0.72,  and 1.2 mg/L.

| 4For a discussion of the precision  of data  from  chronic  toxicity
   tests see Section 4, Quality Assurance.
                                   300

-------
                                                                         'I?-1
     Figure  11.   Data sheet for Champia  parvula sexual  reproduction  test,
                  Receiving  water summary sheet.
                      SITE
COLLECTION DATE '•
TEST DATE
LOCATION












INITIAL
SALINITY










FINAL
SALINITY










SOURCE OF SALTS FOR
SALINITY ADJUSTMENT*










       Ve. natural seawater, GP2 brine, GP2 salts, etc.
       (include some indication of amount)

       COMMENTS:
LFrom Thursby  and Steele,  1987.
                                         301

-------
 Figure 12.  Data  sheet for Champla parvula sexual  reproduction test
             Cystocarp data sheet.
  COLLECTION DATE
   EXPOSURE BEGAN (date)
RECOVERY BEGAN (date)_

COUNTED (date)	
   EFFLUENT OR TOXICANT
                TREATMENTS (% EFFLUENT, ^G/L, or REC. WATER SITES)
888
•


BHBI
•
•

1
i
0

1
I
I
I
IS
B9E
H
| REPLICATES

A 1
2
3
4
MEAN
;
B 1
2
3
4
MEAN
CONTROL


























































































I ' • "• ~
C 1
2
3
4
MEAN











OVERALL
MEAN

Temperature
Salinity
Light




































^Source of Dilution Water
f 5e



|rom Thursby and Steele, 1987
                                     302

-------
                              SELECTED REFERENCES
Anderson, J. W., J. M. Neff, B. A. Cox, H. E. Tatem, and G.  H.  Hightower.
    1974.  Characteristics of dispersions and water-soluble extracts of crude
    and refined oils and their toxicity to estuarine crustaceans and fish.
    Mar. Biol. 27:75-88.

Anonymous.  1979.  Test 6: Mysidopsis bahia life cycle. Fed. Reg.
    44(53):16291.

APHA.   1985.  Standard Methods for the Examination of Water and
    Wastewater.  16th Ed. American Public Health Association, Washington, D.C.

ASTM.   1987.  Standard Guide for Conducting Life Cycle Toxicity Tests
    with Salt Water Mysids.  ASTM E-1191-87,  American Society for Testing and
    Materials, Philadelphia, Pennsylvania.

Bartlett, M. S.  1937.  Some examples of  statistical methods of research
    in  agriculture and applied biology.   J. Royal Statist. Soc. Suppl.
    4:137-183.

Bay, S. M., P. S. Oshida, and K. D. Jenkins.  1983.  A simple new bioassay
    based on echinochrome synthesis by larval sea urchins.  'Mar. Environ. Res.
    8:29-39.

Beck, A. D., and D. A. Bengtson.  1982.   International study on Artemia
    XXII:   Nutrition  in aquatic toxicology  - Diet quality of geographical
    strains of the brine  shrimp, Artemia.   In:  Pearson, J.G., R.B. Foster, and
    W.E. Bishop, eds., ASTM  STP 766, American Society for Testing and
    Materials, Philadelphia, Pennsylvania,  pp.  161-169.

Beck, A. D., D. A. Bengtson, and W. H. Howell.  1980.   International  study
    on  Artemia. V. Nutritional value of  five geographical strains of Artemia:
    Effects of survival and  growth of  larval Atlantic  silversides, Menidia
    menidia.   In:  Persoone,  G., P. Sorgeloos, D. A. Roels,  and E. Jaspers.
    eds., The  brine shrimp,  Artemia. Vol. 3, Ecology, culturing, .use  in
    aquaculture.   Universa  Press, Wetteren,  Belgium,   pp. 249-259.

Beckett, D. C.,  and P.  A.  Lewis.   1982.   An efficient  procedure  for  slide
    mounting  of  larval  chironomids.  Trans.  Amer. Fish. Soc. 101(1}:96-99.

Bengtson, D.  A.   1984.  Resource  partitioning by Menidia menidia and
    Menidia beryllina (Osteichthyes: Atherinidae).  Mar. Ecol. Prog.  Ser.
     18:21-32.
                                       303

-------
 Bengtson, D, A. S., A. D^Beck, S. M. Lussier, D. Migneault, and C. E.
   Olney.  1984.  International study on Artemia. XXXI.  Nutritional effects
   in toxicity tests: use of different Artemia geographical strains.  In:
   Persoone,  G., E. Jaspers, and C. Claus, eds., Ecotoxicological testing for
   the marine environment, Vol. 2.  State Univ. Ghent and Inst.  Mar. Sci.
   Res., Bredene, Belgium,  pp. 399-417.

 Benoit, D. A., F.  A. Puglisi, and D.  L. Olson.  1982.  A fathead minnow,
   Pimephales promeTas, early life state toxicity test method evaluation and
   exposure to four organic chemicals.  Environ. Pollut. (Series A)
   28:189-197.

 Bigelow, H.  B., and W. C. Schroeder.   1953.   Fishes of the Gulf of  Maine.    "
   U. S.  Fish Wild!. Serv., Fish.  Bull. 53:1-577.

 Birge,  W. J., and  J. A. Black.   1981.  jn situ acute/chronic toxicological
   monitoring of industrial  effluents  for the NPDES biomonitoring program
   using fish and amphibian  embryo/larval stages as test organisms.   Office
   of Water Enforcement and  Permits, U.  S. Environmental Protection  Agency,
   Washington, D,C.   OWEP-82-001.

 Birge,  W. J., J. A. Black,  and  B.  A.  Ramey.   1981.   The reproductive
   toxicology of aquatic contaminants.   Hazard assessments  of chemicals,
   current developments, Vol.  1, Academic Press,  Inc.,  p. 59-114.

 Birge,  W.  J.,  J. A. Black,  and  A,  G.  Westerman.   1979.   Evaluation  of  aquatic
   pollutants using  fish and amphibian eggs as bioassay organisms.   National
   Academy of Sciences,  Washington, D.C.,  p.  108-118.

 Birge,  W.  J.,  J. A. Black,  and  A.  G.  Westerman.   1985.   Short-term  fish  and
   amphibian  embryo-larval tests for determining  the  effects  of  toxicant
   stress  on  early  life  stages and  estimating  chronic values  for  single
   compounds  and complex effluents.  Environ.  Tox. Chem.  (4):807-821.

 Birge,  W.  J.,  J. A., Black, A.  G., Westerman,  and B. A. Ramey.   1983.
   Fish  and amphibian embryos - A model  system for evaluating
   teratogenicity.   Fundam.  Appl. Toxicol. 3:  237-242.

 Birge,  W. J., and R. A. Cassidy.   1983.   Importance of  structure-activity
   relationships in  aquatic  toxicology.   Fundam. Appl. Toxicol. 3:359-368.

 Black,  J. A., W. J. Birge, A. G. Westerman, and P. C. Francis.  1983.
   Comparative aquatic toxicology of aromatic  hydrocarbons.   Fundam.  Appl.
   Toxicol. 3: 353-358.

Blaxter, J.  H. S.,   F.  S. Russell, and M. Yonge, eds.   1980.  The
   biology of mysids and euphausiids.  Part 1.  The biology of the  mvsids.
  Adv. Mar.  Biol. 18:1-319.

Borthwick, P. W., J. M. Patrick, Jr.,  and D.  P. Middaugh.  1985.  Comparative
  acute sensitivities of early life stages of atherinid fishes to
  chloropyrifos and thiobencarb.  Arch.  Environm. Contam. Toxicol.
   14:465-473,
                                      304

-------
fBower,  C.  E.   1983.   The basic marine aquarium.   Charles  C.  Thomas,  Pub!.,
   Springfield,  Illinois.

 Ireteler,  R.J.,  O.W.  Williams, and R.L.  Buhl.   1982.   Measurement  of
   chronic  toxicity using the opossum shrimp Mysidopsis bahia.   Hydrobiol.
   93:189-194.
^uikema,  A.  L.,  Jr.,  B.  R.  Niederlehner,  and 0.  Cairns, Jr.   1981.   The
   effects  of simulated refinery effluent  and its components  on  the estuarine
   crustacean,  Mysidopsis bahia.  Arch.  Environm. Contam.  Toxicol.
   10:231-240.
iuikema,  A.  L.,  B.  R. Niederlehner,  and J.  Cairns.   1982.   Biological
   monitoring.   Part IV.   Toxicity Testing.   Water Res.  16:239-262.

 )hernoff, B.,  J. V. Conner,  and C. F.  Byran.  1981.   Systematics of the
   Menidia beryl!ina complex  (Pices:Atherinidae)  from the Gulf of Mexico and
   its tributaries.  Copeia 2:319-335.

[conover,  W.  J.  1980.  Practical nonparametric statistics.   Second edition.
   John Wiley and Sons, New York. pp. 466-467.

 ;ripe, G. M.,  D. R. Nimmo, and T. L. Hamaker.  1981.   Effects of two
   organophosphate pesticides on swimming stamina of  the mysid Mysidopsis
   bahia.   In:  Vernberg,  F. J., A. Calabrese, F.  P.  Thurberg, and W. B.
   Vernberg,  eds.5  Biological monitoring of marine pollutants.  Academic
   Press,  New York,  pp 21-36.

 Javey, E. W.,  J. H. Gentile, S. J. Erickson, and P.  Betzer.  1970.
   Removal of trace metals from marine culture media.  Limnol. Oceanogr.
   15:486-488.

 )eWoskin, R. S.  1984.  Good laboratory practice regulations: a comparison.
   Research Triangle Institute, Research Triangle Park,  N.  Carolina. 63 pp.

Ibinnel, P.,  Q. Stober, J. Link, M. Letourneau, W. Roberts,  S. Felton,  and
   R. Nakatani.  1983.  Methodology and validation of a sperm cell toxicity
   test for testing toxic substances in marine waters.  Final Report. Grant
   R/TOX.  FRI-UW-83.  University of Washington Sea Grant Program in
   cooperation with U.S. Environmental Protection Agency.  208 pp.

[Oixon, W. J.,  and F. J. Massey, Jr.  1983.   Introduction to statistical
   analysis.  Fourth edition.   McGraw Hill, New York.

rDowns, T. R.,  and W. W. Wilfinger.  1983.  Fluorometric quantification of
   DNA in cells and tissue.  Anal. Biochem.  131:  538-547.

 Draper, N. R., and J. A. John.  1981.  Influential observations and outliers
   in regression.  Technometrics 23:21-26.

'Dunnett, C.  W.  1955.  Multiple comparison procedure for comparing several
   treatments with a control.  J. Amer. Statist.  Assoc. 50:1096-1121.
                                       305

-------
 Dunnett, C. W.  1964.  New table for multiple comparisons with a control.
 1  Biometrics 20:482.

 Dye, J. E.  1980.  The production and efficient use of freshly hatched brine
   shrimp nauplii (Artemia) in the larval rearing of marine fish at the
   hatcheries of the British White Fish Authority.  In: Persoone, G., P.
   Sorgeloos, D. A.  Reels, and E. Jaspers. eds.9 The brine shrimp, Artemia.
   Vol. 3, Ecology,  culturing, use in aquaculture.  Universe Press, Wetteren,
   Belgium,  pp. 271-276.

|Farrell, D. H. 1979.  Guide to the shallow-water mysids from Florida.
   Fla. Dept. Environ. Reg., Techn. Ser. 4{l):]-69.                        "|

 :DA.  1978.  Good laboratory practices for nonclinical laboratory studies.
   Part 58, Fed. Reg. 43(247):60013-60020, December  22, 1978.

[finney, D. J.   1948.  The Fisher-Yates test of significance in 2X2
   contingency  tables.  Biometrika 35:145-156.

IFinney, D. J.   1971.  Probit analysis.  Third Edition.   Cambridge Press,
   New York. 668 pp.
re
jFotheringham,  N., and S. L. Brunenmeister.   1975.  Common marine
   invertebrates of  the northwestern  Gulf coast.  Gulf  PubK  Co.,  Houston,
   Texas.

jFujita, S., T.  Watanabe, and C.  Kitajima.  1980.  Nutritional  quality of
   Artemia from different localities  as  a living feed for marine fish from
   the viewpoint of  essential  fatty acids.  In:  Persoone,  G.,  P.  Sorgeloos,
   D. A. Roels,  and  E.  Jaspers,  eds.,  The brine  shrimp,  Artemia.  Vol. 3,
   Ecology,  culturing,  use in aquaculture.  Universa Press,  Wetteren,
   Belgium,   pp.  277-290.
 Sast,  M.  H.,  and  W.  A.  Brungs.   1973.   A procedure  for  separating  eggs  of
   the  fathead minnow.   Prog.  Fish.  Cult.  35:54.

[Gentile,  J. H., S. M. Gentile,  N. G. Hairston, Jr.,  and B. K. Sullivan.
   1982.   The  use  of  life-tables for evaluating the  chronic toxicity of
   pollutants  to Mysidopsis  bahia.   Hydrobiol. 93:179-187.

 ientile,  J. H., S. M. Gentile,  G. Hoffman,  J. F. Heltshe, and H. G. Hairston,
   Jr.   1983.   The effects of  chronic mercury expsoure on the survival,
   reproduction and population dynamics  of Mysidopsis bahia.  Environ.
   Toxicol. Chem.  2:61-68.

[Gentile,  S. M., J. H. Gentile,  J. Walker, and J. F.  Heltshe.  1982.  Chronic
   effects of  cadmium on two species of  mysid shrimp: Mysidopsis bahia and M.
   bigelowi.   Hydrobiol. 93:195-204.
                                      306

-------
.Goodman,  L.  R.,  D.  J.  Hansen,  D.  P. Middaugh,  G. M.  Cripe, and  0. C. Moore.
   1985.   Method  for early  life-stage  toxicity  tests  using three atherinid
   fishes  and results with  chloropyrifos.   In:   Cardwell, R. D., R. Purdy,
   and  R.  C.  Bahner, eds.,  Aquatic Toxicology and Hazard Evaluations ASTM STP
   854, American  Society for  Testing and Materials, Philadelphia,
   Pennsylvania,  pp. 145-154.

[Hansen, D. J., P. R. Parrish,  S.  C. Schimmel,  and L. R. Goodman.  1978.
   Life-cycle toxicity  test using  sheepshead minnows  (Cyprinodon
   variegatus).   In: Bioassay procedures for the ocean disposal  permit
   program, U. S. Environmental  Protection  Agency, Environmental Research
   Laboratory, Gulf  Breeze, Florida, EPA/600/9-78/010,  pp. 109-117.

 teard, R. W.  1982. Guide to  the common tidal  marsh invertebrates of the
   northeastern Gulf of Mexico.  Publ.  No.  MASGP-79-004, Mississippi-Alabama
   Sea  Grant  Consortium,  Ocean  Springs, Mississippi,

Heber, M. A., M. M. Hughes,  S.  C. Schimmel, and D. A. Bengtson.   1987.
   Guidance manual for  rapid  chronic toxicity tests on effluents and
   receiving  waters  with  larval  inland  silversides (Menidia beryllina).
   Contribution No.  792.  In:   Schimmel, S. C.  ed., Users guide  to the
   conduct and interpretation of complex effluent toxicity tests at
   eastuarine/marine sites.   Environmental  Research Laboratory,  U. S.
   Environmental  Protection Agency, Narragansett, Rhode Island.  Contribution
   No.  796.,  265  pp.

Hildebrand,  S. F.   1922.   Notes on habits  and  development of eggs and larvae
   of the  silversides Menidia menidia and Henidia beryllina.  Bull. U. S.
   Bur. Fish.  38:113-120.
Hildebrand,  S. F.,  and W. C. Schroeder.
   Bull. U. S. Fish. 43(l):366pp.
1928.   Fishes of Chesapeake Bay.
Hodges,  J. L., Jr.,  and  E. L. Lehmann.   1956.  The efficiency of some
   nonparametric competitors of the t-test.  Ann. Math. Statist. 31:625-642,

Home, J. D., M. A.  Swirsky, T. A. Hollister, B. R. Oblad, and J. H.
   Kennedy.   1983.  Aquatic toxicity studies of five priority pollutants,
   Contract Report, EPA Contract No. 68-01-6201, U. S. Environmental
   Protection Agency, Washington, DC.

Horning, W.  B., and  C. I. Weber.  1985.  Short-term methods for estimating
   the chronic toxicity of effluents and  receiving waters to freshwater
   organisms.  Environmental Monitoring and Support Laboratory, U. S.
   Environmental Protection Agency, Cincinnati, Ohio. 45268,
   EPA/600/4-85/014.
                                      307

-------
Hughes,  M.  M.,  M.  A.  Heber,  S.  C.  Schimmel,  and  U.  J. Berry.   1987.   Guidance
   manual  for  conducting  complex effluent  and receiving water  larval fish
   growth-survival  studies  with  the sheepshead minnow  (Cyprinodon
   variegatus).   Contribution No.  X104.   In:   Schimmel, S. C.,  ed.  Users
   guide  to  the  conduct and interpretation of complex  effluent  toxicity tests
   at  eastuarine/marine sites.   Environmental  Research Laboratory,  U.  S.
   Environmental  Protection Agency,  Narragansett,  Rhode Island.  Contribution
   No.  796., 265  pp.

Hutton,  C.  H.,  P.  F.  DeLisle, M.  H.  Roberts,  and  D. A. Hepworth.   1986.
   Chrysaora quinguecirrha: a predator on  mysids  (Mysidopsis bahia) in
   culture.  Progr.  Fish-Cult. 48:154-155.

Jackim,  E., and  D.  Nacci.   1986.   Improved sea urchin DNA-based embryo growth
   toxicity  test.   Environ. Toxicol.  Chem.   5:561-565.

Jensen,  A.  1984a.  Marine ecotoxicological  tests with seaweeds.
   In:  Personne,  G., E. Jaspers,  and  C.  Claus, eds., Ecotoxicological  testing
   for  the marine environment. Vol.  1.   State University of Ghent and
   Institute for  Marine Scientific  Research,  Bredene,  Belgium,  pp. 181-193.

Jensen,  A.  1984b.  Marine ecotoxicological  tests with phytoplankton.
   In:  Personne,  G., E. Jaspers,  and  C.  Claus, eds., Ecotoxicological  testing
   for  the marine environment. Vol.  1.   State University of Ghent and
   Institute for  Marine Scientific  Research,  Bredene,  Belgium,  pp. 195-213.

Jensen,  J.  P.   1958.  The  relation between body  size  and number of eggs in
   marine malacostrakes.  Meddr.  Danm. Fisk.-og Havunders 2:1-25.

Johns, D. M., W. J. Berry, and  W.  Walton.   1981.  International study on
   Artemia.  XVI.  Survival,  growth  and reproductive potential of the mysid
   Mysidopsis  bahia  Molenock  fed various geographical  strains of the brine
   shrimp Artemia.   J. Exp. Mar.  Biol. Ecol.  53:209-219.

Johns, D. M., M. E. Peters,  and A.  D. Beck.   1980.  International  study on
   Artemia.  VI.   Nutritional  value  of geographical and temporal strains of
   Artemia:.effects  on survival  and growth  of two  species of Brachyuran .
   larvae.   In:  Persoone, G.,  P.  Sorgeloos,  D. A.  Roels, and E. Jaspers.
   eds.,  The brine  shrimp,  Artemia.  Vol. 3,  Ecology, culturing, use in
   aquaculture.   Universa Press,  Wetteren,  Belgium,  pp. 291-304.

;Johnson,  M. S.   1975.  Biochemical  systematics of the atherinid genus
   Menidia.  Gopeia  4:662-691.

.Kenaga,  E.  E.   1982.  The  use of  environmental toxicology and chemistry data
   in  hazard assessment:  Progress,  needs,  challenges.  Environ. Toxicol.
   Chem.  1:69-79.

Kenaga,  E.  E.,  and  R. J. Moolenaar.  1976.   Fish  and  Daphnia toxicity as
   surrogates  for aquatic vascular  plants  and algae.   Environ. Sci. Techno!.
   13:1479-1480.
                                      308

-------
 Jester,  D.  R.,  I.  W. Dredall,  D.  N.  Connors,  and R.  M.  Pytokowicz.   1967^1
   Preparation of artificial  seawater.   Limnol.  Oceanogr.  12:176-179.

 :iein-MacPhee,  G.  W. H.  Howell,  and  A.  D.  Beck.   1980.   International  study
   on Arterma.  VII.   Nutritional  value of five geographical  strains  of
   Artemia ^ winter flounder Pseudopleuronectes  americanus  larvae.    In-
   Persoone,  G.,  P.  Sorgeloos,  D.  A.  Roels,  and  E.  Jaspers.  eds.,  The brine
   shrimp, Arterma.  Vol.  3, Ecology,  culturing,  use in aquaculture.  Universa
   Press,  Wetteren,  Belgium,   pp.  305-312.

 ilein-MacPhee,  6.,  W.  H.  Howell,  and A.  D.  Beck.   1982.   International  study
   on Artercna. XX.   Comparison  of  a reference  and four geographical  strains
   of Arterma as  food for  Winter flounder (Pseudopleuronectes americanus)
   larvae.  Aquacult. 29:279-288.	.;.,;

 U>pp,_J.  F.   1983.   Guidelines and format for EMSL-Cincinnati methods.    ''?
   Environmental Monitoring and Support Laboratory, U. S.  Environmental
   Protection Agency, Cincinnati,  Ohio,   EPA/600/8-83/020.

 iuntz, A.  1916.  Notes on the embryology and larval development of five
   species of teleostean fishes.   Bull. U. S. Bur. Fish.  34(831):409-429.

 .awler, A. R., and S.  L.  Shepard.  1978.  Procedures for eradication of
   nydrozoan pests in closed-system mysid culture.  Gulf Res. Rept. 6:177-178.

 -eger, P., and P. Sorgeloos.   1982.  Automation in stock-culture maintenance
   and juvenile separation of the mysid Mysidopsis bahia  (Molenock).
  Aquacult.  Eng. 1:45-53.                     	

 .eger, P., Sorgeloos, P., Millarnena,  0.  M., and  Simpson, K.  L.   1985.
   International Study on Artemia.  XXV.   Factors determining the
  nutritional effectiveness of Artemia:   The relative impact of chlorinated
  hydrocarbons and essential  fatty acids in San  Francisco Bay and San  Pablo
  Bay Artemia."   J.  Exp.  Mar. Biol. Ecol.,  Vol.  93, 1985.  pp.  71-82.

 [eger, P., Bengtson, D. A,,  Simpson,  K.  L.  and Sorgeloos,  P.  1986.   "The
  use and nutritional value of Artemia as a food source,"  M. Barnes  (ed )
  Oceanography and  MajMne Biology AnmiaJ £gvipw3  Vol.  24,  Aberdeen     ' '
  University Press,  Aberdeen, Scotland,  1986,  pp.  521-623.

[ussier,  S.  M.,  J.  H. Gentile,  and J. Walker.  1985.   Acute  and  chronic
  effects of  heavy  metals  and cyanide on  F-tysidopsIs bahia  (Crustacea:
  Mysidacea).  Aquat. Toxicol.  7:25-35.

[ussier,  S. M.s  A. Kuhn,  and  J. Sewall.   1987.  Guidance manual  for
  conducting  seven day  mysid  survival/growth/reproduction  study  using the
  estuarine mysid, Mysidopsis bahia.  Contribution  No. X106.  In:  Schimmel,
  S.  C.,  ed.   Users  guide  to  the conduct  and interpretation  of complex
  effluent toxicity  tests  at  estuarine/marine  sites.  Environmental  Research
 Laboratory, U. S.  Environmental  Protection Agency, Narragansett, Rhode
  Island.  Contribution No. 796.,  265 pp.
                                     309

-------
     rtacek, K. J.,  and B. H. Sleight.  1977.   Utility of toxicity tests with
       embryos and  fry of fish in evaluating  hazards associated  with  the chronic
       toxicity of  chemicals to fishes.   In:   Mayer, F.  L.,  and  J.  L.  Hamelink,
       eds.s  Aquatic Toxicology and Hazard Evaluation,  ASTM  STP  634,  American
       Society for  Testing and Materials,  Philadelphia,  Pennsylvania,  pp.  137-146,

     Mauchline,  J.,  and M. Murano.   1977.   World  list of the Mysidacea,
       Crustacea.   J.  Tokyo Univ.  Fish.  64:39-88.

     McKim, J.  M.   1977.   Evaluation of  tests  with  the  early life stages of
       fish for  predicting long-term toxicity.  J,  Fish.  Res.  Bd.  Can.
       34:1148-1154.

     Middaugh,  D. P.,  and M.  J.  Hemmer.   1984.  Spawning  of  the  tidewater
       silverside, Menidia peninsulae (Goode and  Bean)  in response  to  tidal  and
       lighting  schedules in the laboratory.   Estuar. 7:137-146.

     Middaugh, D. P.,  M.  J.  Hemmer,  and  L,  R.  Goodman.   1987.  Methods  for
       spawning, culturing and  conducting  toxicity-tests with  early life stages
       of four antherinid fishes: the inland silverside, Menidia  beryllina,
       Atlantic  silverside,  M.  menidia,  tidewater silverside,  M.  peninsulae, and
       California grunion,  Leuresthes tenuis.  Office of Research and
       Development,  U.  S.  Environmental  Protection Agency, Washington,  DC 20460,
       EPA/600/8-87/004.

     Middaugh, D. P.,  M.  J.  Hemmer,  and  Y.  Lamadrid-Rose.  1986.  Laboratory
       spawning cues in Menidia  beryllfna and M. pennisulae  (Pices, Atherinidae)
       with notes on survival and growth of larvae at different salinities.
       Environ. Biol.  Fish.  15(2):107-117.

     Middaugh, D. P.,  and  T.  Takita.   1983.  Tidal and diurnal  spawning  cues in
       the  Atlantic Silverside, Menidia menidia.  Environ. Biol. Fish.
       8(2):97-104.              	'	
    Miller, R. G.  1981.
      New York. 299 pp.
Simultaneous statistical  inference.   Springer-Verlag,
SfrfS;
if,  Molenock, J,  1969.  Mysidopsis bahia, a new species of mysid (Crustacea:
      Mysidacea) from Galveston Bay, Texas.  Tulane Stud. Zool. Bot.
    Morgan, M.D.  1982.  The ecology of Mysidacea.  Developments in
      hydrobiology 10, W. Junk, Pub!., The Hague, Netherlands.   232 pp.

    Mount, D. I., and T. Norberg-King.  1986.  Validity of effluent and  ambient
      toxicity tests for predicting biological  impact,  Kanawha  River,
      Charleston, West Virginia.  Environmental Research Laboratory,  U.  S.
      Environmental  Protection Agency, Duluth,  Minnesota,  EPA/600/3-86/006.

    Mount, D. I., A. E. Steen, and T.  Norberg-King.   1986.   Validity  of  effluent
      and ambient toxicity tests for predicting biological  impact,  Back  River,
      Baltimore Harbor, Maryland.   Environmental  Research  Laboratory,  U.  S.
      Environmental  Protection Agency, Duluth,  Minnesota,  EPA/600/8-86/001.

                                         310

-------
 lount,  D.  I.,  T.  Norberg-King,  and  A.  E.  Steen,   1986.   Validity of  effluent
   and ambient  toxicity tests  for  predicting  biological  impact,  Naugatuck
   River, Connecticut.   Environmental Research  Laboratory, U. S.
   Environmental Protection  Agency,  Duluth, Minnesota, EPA/600/8-86/005.

 lount,  D.  I.,  and T. J.  Norberg-King,  eds.   1985.  Validity of  effluent and
   ambient  toxicity tests for  predicting biological impact, Scippo Creek,
   Circleville, Ohio.   Environmental Research Laboratory, U. S.  Environmental
   Protection Agency, Duluth,  Minnesota, EPA/600/3-85/044.

 |ount,  D.  I., A.  E. Steen,  and  T. J. Norberg-King, eds.  1985.  Validity of
   effluent and ambient toxicity testing for predicting  biological impact on
   Five  Mile Creek,  Birmingham,  Alabama.   Environmental  Research Laboratory,
   U. S. Environmental  Protection  Agency,  Duluth, Minnesota.
   EPA/600/8-85/015.

jflount,  D.  I.,  , A.  E.  Steen,  and  T. J. Norberg-King, eds.  1985.  Validity
   of effluent and  ambient toxicity tests  for predicting biological impact,
   Ohio  River, near Wheeling,  West Virginia.  Environmental Research
   Laboratory, U. S. Environmental Protection Agency, Duluth, Minnesota.
   EPA/600/3-85/071.

 lount,  D.  I., and  C. E.  Stephan.  1967.   A method for establishing acceptable
   toxicant limits  for  fish  -  Malathion and 2,4-D.  Trans. Am.  Fish. Soc. 96:
   185-193.

 fount,  D.  I., N. A. Thomas, T.  J. Norberg, M. T, Barbour, T.  H. Roush,
  and W. F. Brandes.   1984.   Effluent and ambient toxicity testing and
   instream community response on the Ottawa River, Lima, Ohio.
  Environmental Research Laboratory, U. S. Environmental Protection Agency,
  Duluth, Minnesota, EPA/600/3-84/080.

  cci, D.,  and E.   Jackim.   1985.  Rapid aquatic toxicity assay using
   incorporation of tritiated-thymidine into sea urchin,  Arbacia punctulata,
  embryo.  In: Bahner,  R. C., and D. J. Hansen. eds., Aquatic  Toxicology and
  ^Hazard Assessment: Eighth Symposium,   STP 891. American Society for
  Testing and Materials, Philadelphia,  Pennsylvania,  pp. 382-393.
  cci, D., E. Jackim, and R. Walsh.  1986.  Comparative evaluation of three
  t apid marine toxicity tests: Sea urchin early embryo growth test, sea
  :urchin sperm cell toxicity test and Microtox.  Environ.  Toxicol. Chem.
  5:521-525.
«»>
gcci, D., R. Walsh, and E.  Jackim.  1987.  Guidance manual  for conducting
 'sperm cell tests with the  sea urchin,  Arbacia punctulata,  for use in
  testing complex effluents.  Contribution No.  X105.  In:   Schimmel,  S. C.,
  ed.   Users guide to the conduct and interpretation of complex effluent
  toxicity tests at estuarine/marine sites.   Environmental  Research.
  Laboratory, U. S. Environmental Protection Agency, Narragansett, Rhode
  Island.  Contribution No.  796., 265 pp.
                                     311

-------
 immo, D. R,,  L.  H. Banner, R. A. Rigby, J. M. Sheppard, and A. J.
  Wilson, Jr.   1977.  Mysidopsis faahi'a: an estuarine species suitable for
  life-cycle toxicity tests to determine the effects of a pollutant.  _In:
  Mayer, F. L., and J. L. Hamelin, eds,, Aquatic Toxicology and Hazard
  Evaluation.   ASTM STP 634, American Society for Testing and Materials,
  Philadelphia, Pennsylvania, pp. 109-116.
  [mmo, D. R., and T. L. Hamaker.  1982.
  freview.  Hydrobiol. 93:171-178.
Mysids in toxicity testing - a
 SB
 |jirno9 D. R.5 T. L. Hamaker, J. C. Moore, and C. A. Sommers.   1979.  Effect
   |f diflubenzuron on an estuarine crustacean.  Bull. Environm. Contam.
   foxicol. 22 :767-770.

   nmo, D. R., T. L. Hamaker, E. Matthews, and J. C. Moore. 1981.  An
   Overview  of the acute and chronic effects of first and second generation
   >esticides on an estuarine mysid.  In: Vernberg, F. J., A.  Calabrese, F.
     Thurberg, and W.B. Vernberg, eds.,  Biological Monitoring of Marine
   }ollutants.  Academic Press, New York. pp. 3-19.

   imo, D. R., T. L. Hamaker, E. Matthews, and W. T. Young.  1982.  The long-
  term effects of suspended particulates on survival and reproduction of the
 |ysid shrimp,  Mysidopsis bahia, in the laboratory.  In: Mayer, G. F., ed.,
   cological Stress and the New York Bight: Science and Management.
   stuarine Res. Found., Columbia, S.  Carolina,  pp. 41-50.
          R., T. 1. Hamaker,  J.  C.  Moore,  and R.  A,  Wood.   1980.   Acute
   nd chronic effects of Dimilin on survival  and  reproduction  of  Mysidopsis
   ahia.   In: Eaton, J.  G.,  P.  R. Parrish,  and A.C.  Hendricks,  eds.,  ASTM
  |TP 707,  American Society for Testing and Materials,  Philadelphia,
   ennsylvania.  pp. 366-376.

 pio,  D.  R.,  T.  L. Hamaker,  and C.  A.  Sommers.   1978a.   Culturing  the  mysid
 fMysidopsis  bahia)  in flowing sea  water  or a static system.   In:  Bioassay
 "rocedures for  the Ocean Disposal  Permit Program,  U.S.  Environmental
  'rotection Agency,  Environmental Research Laboratory, Gulf Breeze,
   lorida,  EPA/600/9-78/010,  pp.  59-60.
       D.  R.,  T.  L.  Hamaker,  and  C.  A.  Sommers.   1978b.  Entire  life cycle
 toxicity test  using  mysids  (Mysidopsis  bahia)  in  flowing water.   In:
 Bioassay Procedures  for  the Ocean  Disposal  Permit Program, U.S.
 Environmental  Protection Agency, Environmental  Research Laboratory, Gulf
 Breeze,  Florida, EPA/600/9-78/010,  pp.  64-68.

Ijmmo,  D.  R.,  and E. S.  Iley,  Jr.  1982  Culturing  and chronic toxicity of
 Mysidopsis bahia using artificial  seawater. Office of Toxic Substances,
 U.S.  Environmental Protection Agency, Washington,  DC.,.Publ. PA 902.

limmo,  D.  R.,  R. A.  Rigby,  L.  H.  Bahner,  and  J. M.  Sheppard.  1978.  The
 acute and chronic  effects of cadmium  on the estuarine mysid, Mysidopsis
 bahia.   Bull. Environm.  Contam. Toxicol.  19(l):80-85.
                                     312

-------
 Norberg, T, J., and D. I. Mount.  1983.  A seven-day larval growth test.
   Presented at the Annual Meeting, Society of Environmental Toxicology and
   Chemistry, November 6-9, 1983, Arlington, Virginia.

 jNorberg, T. J., and D. I. Mount.  1985.  A new fathead minnow
   (Pimephales promelas) subchronic toxicity test.  Environ. Toxicol.
   Chem. 4(5):711-718.

 [orberg, T. J., and D. I. Mount, eds.  1986  Validity of effluent and ambient
   toxicity tests for predicting biological impacts Skeleton Creek, Enid,
   Oklahoma.  Environmental Research Laboratory,  U. S. Environmental
 I Protection Agency, Duluth,  Minnesota, EPA/600/8-86/002.
 parson, E.  S., and T. 0, Hartley.  1962.  Biometrika tables for
   statisticians.  Vol. 1. Cambridge Univ. Press, England,  pp. 65-70.

 HItier, W.   1978.  Methods for measuring the acute toxicity of effluents
   to aquatic organisms.  Second edition.   Environmental  Monitoring and
   Support Laboratory - Cincinnati, U.  S.  Environmental Protection Agency,
   Cincinnati, Ohio, EPA/600/4-78/012.

 6eltier, W., and C. I. Weber,  eds.  1985.  Methods for measuring the  acute
   toxicity of effluents to freshwater  and marine organisms.   Third edition.
   Environmental Monitoring and Support Laboratory, U.  S. Environmental
   Protection Agency, Cincinnati,  Ohio, EPA/600/4-85/013. 230 pp.

 Jersoone, G., E. Jaspers, and  C.  Claus, eds.   1980.  Ecotoxicological  testing
   for the marine environment.  Vol. 1.  State University of  Ghent and
   Institute  for Marine Scientific Research, Bredenes
 'ersoone,  G.,  P.  Sorgeloos,  0.  Roels,  and E.  Jaspers,  eds.   1980a.  The
   brine shrimp Artemia.   Vol.  1.   Morphology,  genetics,  radiobiology,
   toxicology.   Proceedings  of  the International  Symposium on  the  brine
   shrimp Artemia  salina,  Corpus Christi,  Texas,  1979.  Universa Press,
   Wetteren,  Belgium.  318  pp.
-Persoone,  G.,  P.  Sorgeloos,  0.  Roels,  and  E.  Jaspers,  eds.   1980b.  The
   brine  shrimp Artemia.   Vol. 2.   Physiology,  biochemistry,  molecular
   biology.   Proceedings  of the  International  Symposium on the brine shrimp
   Artemia  salina,  Corpus  Christi,  Texas, 1979.   Universa Press, Wetteren,
   Belgium.  636 pp.

^Persoone,  G.,  P.  Sorgeloos,  0.  Roels,  and  E.  Jaspers,  eds.   1980c.  The
   brine  shrimp Artemia.   Vol. 3.   Ecology, culturing,  use in aquaculture.
   Proceedings  of  the  International Symposium  on  the brine shrimp Artemia
   salina,  Corpus  Christi, Texas, 1979.  Universa Press, Wetteren, Belgium.
   428 pp.

Price, W.  W.   1978.   Occurrence of Mysidopsis  almyra Bowman, M. bahia
   Molenock,  and Bowmaniella  brasiliensls Bacescu (Crustacea, Mysidacea) from
   the eastern  coast of Mexico^Gulf Res.  Repts. 6:173-175.
                                      313

-------
 'rice,  W.  W.   1982.   Key to  the  shallow water  Mysidacea of  the  Texas  coast
   with  notes  on  their ecology.   Hydrobiol.  93(1/2):9-21.
(jRand,  G.  M.,  and  S.  R.  Petrocelli.   1985   Fundamentals  of aquatic
   toxicology.   Hemisphere  Publ.  Corp.,  New York.  66  pp.

|ieitsemas  L.  A,   1981.   The  growth,  respiration,  and energetics of
   Mysidopsis  almyra  (Crustacea;  Mysidacea)  in relation  to temperature,
   salinity, and hydrocarbon  exposure. Ph.D. thesis,  Texas A & M University,
   College  Station, Texas.
 [eitsema,  L. A.,  and  J. M.  Neff.   1980.  A recirculating artificial
   seawater system for the  laboratory culture of Mysidopsis almyra
   (Crustacea; Pericaridea).   Estuaries 3:321-323.

 ienfro, W. C.   1960.   Salinity relations of some fishes in the Aransas
   River.   Tulane  Stud. Zool.  8:83-91.

 :ichards,  F. A.,  and  N. Corwin.   1956.  Some oceanographic applications of
   recent determinations of  the solubility of oxygen  in sea water.  Limnol.
   Oceanogr.  1(4):263-267.

 tieder, D.   1985.  Hazard  evaluation division: Standard evaluation
   procedure: Acute toxicity test  for estuarine and marine organisms  (shrimp
   96-hour  acute toxicity test).   Office of Pesticides Programs, U. S.
   Environmental Protection Agency, Washington, DC, 20460, EPA/540/9-85/010.

 lodgers, J.  H., Jr.,  P. B. Dorn,  T. Duke, R. Parrish, and B. Venables
   (rapporteur).   1986.  Mysidopsis sp.: life history and culture.  A report
   from a workshop held at Gulf Breeze, Florida, October 15-16, 1986.
   American Petroleum  Institute, Washington, DC.

 icheffe, H.  1959.  The analysis  of variance.  John Wiley, New York.  477 pp.

 ichimmel, S. C., ed.   1987.   Users guide to the conduct and interpretation of
   complex effluent toxicity tests at estuarine/marine sites.  Environmental
   Research Laboratory, U. S.  Environmental Protection Agency, Narragansett,
   Rhode Island.  Contribution No. 796., 265 pp.

 limmons, E.  G.  1957.  Ecological survey of the Upper Laguna Madre of
   Texas.  Publ. Inst. Mar. Sci. 4(2}:156-200.
 ir
 Smith, B. A.  1971.   An ecological study of the Delaware River in the
   vicinity of Artificial Island.  Part V.  The fish of four low-salinity
   tidal tributaries of the Delaware River estuary.  Progress Report to
   Public Service Electric and Gas Co.  Ichthyological Assoc. Ithaca, N.Y.
   291 pp.

 medecor, G. W., and  W. G. Cochran.  1980.  Statistical Methods.   Seventh
   edition.   Iowa State University Press,  Ames.  593 pp.
                                      314

-------
forgeloos,  P.  1980.   Life  history of  the  brine  shrimp Artemia.   In:
  Persoone,  G.,  P. Sorgeloos,  D.A,  Roels,  and E.  Jaspers,  eds.,  The  brine
  shrimp, Artemia. Vol.  1.   Morphology, genetics,  radiobiology,  toxicology.
  Proceedings  of the  International  Symposium on the  brine  shrimp Artemia
  salina, Corpus Christi,  Texas,  1979.  Universa  Press, Wetteren,  Belgium.
  pp.  ixx-xxii.

|ehar,  R.  L..,  D.  K.  fanner,  and  B. R.  Nordling.   1983.  Toxicity of the
  synthetic  pyrethroids, permethrin,  AC-222,  705,  and their  accumulation in
 fearly life stages of  fathead minnows  and snails.   Aquat. Toxicol.
 [3:171-182.
  otte,  S.
  N.Y.
1973.   Marine aquarium keeping.   John Wiley and  Sons,  New York,
 iotte,  S.,  G.  Adams,  and  P.  M.  Bubucis.   1984.   GP2 as an artificial
  seawater for  culture or  maintenance  of marine organisms.  Zool. Biol.
  3:229-240.

 ;eel, R. G.   1959.  A multiple  comparison  rank sum test: treatments
  versus control.  Biometrics 15:560-572.
teel,  R.  G.,  and  J.  H.  Torrie.   1960.   Principles and procedures of
  statistics with  special  reference  to  biological sciences.  McGraw-Hill
  Publ.,  New York.
£teele,  R.  L.,  and  M.,  D.  Hanisak.   1978.   Sensitivity of some  brown  algal
 reproductive  stages  to  oil  pollution.   In:  Jensen, A. and J. R. Stein,
 eds.,  Proceedings of the  Ninth  International  Seaweed Symposium, Vol. 9,
 Science Press,  Princeton, New Jersey,   pp.  181-190.

 ;eele,  R.  L.,  and  G.  B.  Thursby,   1983.   A toxicity test using  life stages
 • of  Champia  parvula (Rhodophyta).   In: Bishop, W. E., R. D. Cardwell, and
 • B.  B.  Heidolph, eds., Aquatic Toxicology and  Hazard Assessment: Sixth
 •Symposium.  STP 802,  American Society for Testing and Materials,
 ^Philadelphia,  Pennsylvania, pp. 73-89.

 fuck, K. C., H.  M.  Perry,  and R. W. Heard.   1979a.  An annotated key to the
 iMysidacea of  the  North  Central  Gulf of  Mexico.  Gulf Res. Rept.
 [6(3):225-238.

 uck, K. C., H.  M.  Perry,  and R. W. Heard.   1979b.  Records and range
 extensions  of  Mysidacea from coastal and shelf water of the  Eastern Gulf
 of  Mexico.  Gulf  Res. Kept. 6(3}:239-248.
fgatz, M.  E., and D.  L. Dudley.   1961.  Seasonal occurrence of marine
 fishes  in four  shore habitats near Beaufort, N.C.,  1957-1960.  U. S. Fish.
 Wild!.  Serv. Sc. Rept. Fish. 390.  19 pp.

iarzwell, C. M.   1971.  Bioassays  to determine allowable waste concentrations
 in the  aquatic  environment.  In:  Cole, H. A.  (Organizer),  A discussion
 on biological effects of pollution in the  sea.  Proc. Roy. Soc. Lond. B
 177(1048):279-285.                                                         j
                                     315

-------
  "attersall, W. M., and 0. S. Tattersall.
   Royal Soc. London. 460 pp.
1951.   The British  Mysidacea.
  "heilacker, G. H., and M. F. McMaster.  1971.  Mass culture of the rotifer
   Brachionus plicatilis and its evaluation as a food for larval anchovies
   Mar. Biol. 10:183-188.
  i
  Ihompson, R. $., and E. M. Burrows.  1984. ' The toxicity of copper, zinc, and
   mercury to the brown macroalga Laminaria saccharina.  In: Persoone, G.,
   E. Jaspers, and C. Glaus, eds., Ecotoxicological testing for the marine
   environment, Vol. 2, State University of Ghent and Institute for Marine
   Scientific Research, Bredene, Belgium,  pp. 259-269.

  •hursby,  G. B.,  and R. L. Steele.  1984.  Toxicity of arsenite and arsenate
   to the  marine  macroalga Champia parvula.  Environ. Toxicol,  Chem.
   o « 01? i — oy / »

  'hursby,  G. B.,  and R. L. Steele.  1986.  Comparison of short- and
   long-term sexual  reproduction tests with the marine red  alga Champia
          -  Environ.  Toxicol.  Chem. 5:1013-1018.                -
|Thursby,  G.  B.,  and R.  L.  Steele.   1987.   Guidance manual  for  conducting
   sexual  reproduction test with  the marine macroalga  Champia parvula  for  use
   in  testing complex effluents.  Contribution  No.  XIO'3.   In:   Schimmel, S.
   C.^ed.  Users guide  to  the  conduct  and  interpretation  of complex effluent
   toxicity tests at eastuarine/marine  sites.   Environmental Research
   Laboratory,  U.  S. Environmental  Protection Agency,  Narragansett, Rhode
   Island.  Contribution  No.  796.,  265  pp.

 •hursby,  G.  B.,  R.  L. Steele,  and  M. E. Kane.   1985.  Effect of organic
   chemicals  on growth and  reproduction in  the  marine  red alga  Champia
   parvula.   Environ.  Toxicol.  Chem.  4:797-805.                —   —

 1SEPA.  1975.  Methods for  acute toxicity  tests with  fish,
   macroinvertebrates, and  amphibians..   Environmental  Research Laboratory
   U. S. Environmental Protection Agency, Duluth, Minnesota.
   EPA/660/3-75/009.

  iEPA.  1977.  Occupational health and  safety manual.  Office of Planning
   and Management, U. S. Environmental  Protection Agency, Washington,  DC.

 [SEPA.  1979a.  Handbook for analytical quality control in water and
  wastewater laboratories.  U. S. Environmental Protection Agency,
   Environmental Monitoring and Support Laboratory, Cincinnati,  Ohio,
  EPA/600/4-79/019.

 [SEPA.  1979b.  Methods for chemical analysis of water and wastes.
  Environmental Monitoring and Support Laboratory, U.  S. Environmental
  Protection Agency, Cincinnati,  Ohio,  EPA-600/4-79/020, revised March 1983
                                     316

-------
   :USEPA. 1979c.  Interim NPDES compliance biomonitoring inspection manual.
     Office of Water Enforcement, U. S. Environmental Protection Agency,
     Washington, D.C.  (MCD-62).

   j$SEPA.  1979d.  Good laboratory practice standards for health effects.
     Paragraph 772.110-1, Part 772 - Standards for development  of test data.
  *&& Fed. Reg. 44:27362-27375, May 9,  1979.
  «:
  VSfefe
  |USEPA.  1980a.  Appendix B - Guidelines for Deriving Water Quality Criteria
  III for the Protection of Aquatic Life and Its Uses.  Federal  Register,  Vol.
  1SJ45, No. 231, Friday, November 28, 1980.
  gUSEPA.   1980b.   Proposed good laboratory practice guidelines  for  toxicity
    .testing.   Paragraph 163.60-6.   Fed.  Reg.  45:26377-26382,  April  18,  1980.

  gpEPA.   1980c.   Physical, chemical,  persistence,  and  ecological effects
  fc testing;  good laboratory practice  standards  (proposed  rule).  40 CFR  772,
  ^  Fed.  Reg. 45:77353-77365,  November 21,  1980.

  flJSEPA.   1981a.   Acute toxicity test  standard using mysid shrimp in static
   w and flow-through systems.   Toxic Substances  Control  Act,  Section 4.   U.S.
   j Environmental Protection Agency, Office of Toxic Substances, Health and
   I Environmental Review Division,  Washington, DC.  17 pp.

           1981b.   Chronic toxicity  test  standard  using  mysid  shrimp  in a
     flow-through  system.  Toxic Substances  Control  Act,  Section 4.   U.S.
     Environmental Protection. Agency, Office of Toxic Substances, Health and
     Environmental Review Division,  Washington, DC.  19 pp.

  sflJSEPA.   1981c.   Technical support document  for  using  mysid  shrimp  in acute
   ( and chronic toxicity tests.  Toxic Substances Control  Act,  Section 4.
   | U.S.  Environmental  Protection Agency, Office  of Toxic  Substances, Health
   | and Environmental  Review Division, Washington,  DC.  43  pp.

{|8?EPA.   1982*  Hetnods for organic chemical analysis  of  municipal  and
     industrial  wastewater.   Environmental Monitoring and Support Laboratory,
r|jjU.  S.  Environmental  Protection  Agency,  Cincinnati,  Ohio,  EPA/600/4-82/057,
 ;|lW—.   1984.   Development  of water quality-based permit  limitations for
 ^fctoxic  pollutants:  National Policy.  Fed. Reg. 49:9016-9019.

 |J|EPA.   1985.   Technical  Support Document for Water Quality-based Control.
     Office  of Water,  U.  S. Environmental Protection Agency, Washington, D.C.

 fisher, R.R., and  D.A. Bengtson.  1981.  Survival and growth of sheepshead
 ij|nn'nnow larvae and  juveniles on diet of Artemia nauplii. Prog. Fish-Cult.
 K43(2):102-105.

 "^Vanhaecke,  P.  and Sorgeloos, P.  1980.  "International Study on Artemia.
    IV.  The  biometrics of  Artemia strains from different geographical
    origin."  G.  Persoone, P. Sorgeloos, 0. Reels and E. Jaspers (eds.), The
    Brine  Shrimp  Artercia, Vol. 3 Ecology, Culturing Use jrn Aquaculture,
    Universa  Press,  Wetteren, Belgium, 1980, pp. 393-405.

                                        317

-------
 |Vanhaecke, P., Steyaert, H. and Sorgeloos, P.  1980.  "International
   Study on Artemia.  III.  The use of Coulter Counter equipment for the
   biometrical analysis of Artemia cysts.  Methodology and Mathematics,"
   G. Persoone, P. Sorgeloos, 0. Roels, and E. Jaspers {eds.}, The Brine
   Shrimp Artemia, Vol. J, Morphology, Genetics, Radiobiology,
   Toxicology.  Universa Press, Wettersen, Belgium, 1980,  pp. 107-115.

 fanhaecke, P., and P. Sorgeloos.  1983.  International study on Artemia.
   XIX.  Hatching data for ten commercial sources of brine shrimp cysts
   and re-evaluation of the "hatching efficiency" concept.  Aquacult.
   30:43-52.

 Jalsh, G.  E., L. H. Banner, and W. B. Horning.   1980.  Toxicity of textile
   mill effluents to freshwater and estuarine algae, crustaceans, and
   fishes.  Environ. Pollut. (Ser. A.) 21:169-179.

 Walters, D. B.,  and C.  W. Jameson.  1984.  Health and safety for toxicity
   testing.  Butterworth Publ., Woburn, Massachusetts.

 Ward, G. $.,  and P. R.  Parrish.  1980.  Evaluation of early life stage
   toxicity tests with embryos and juveniles of  sheepshead minnows.  In:
   J. G.  Eaton, P. R.  Parrish and A.  C. Hendricks, Aquatic Toxicology,
   ASTM STP 707,  American Society for Testing and Materials,  Philadelphia,
   Pennsylvania,   p. 243-247.

 Ward, S. H.  1984.  A system for laboratory rearing of the mysid,
   Mysidopsis  bahla Molenock.  Progr.  Fish-Cult.  46(3):170-175.

 Ward, S. H.  1987.  Feeding response of the mysid Mysidopsis bahia reared
   on Artemia.  Prog.  Fish-Cult. 49:29-33.

 Weber,  C.  I., ed.  1973.  Biological  field and  laboratory methods  for
   measuring the  quality of surface waters and wastes.  Office of Research
   and Development, U.  S. Environmental Protection Agency,  Cincinnati,
   Ohio,  EPA/670/4-73/001.  200 pp.

 Weber,  C.  I., and W.  H.  Peltier.   1981.   Effluent toxicity screening test
   using  Daphnia  and mysid shrimp.  Environmental  Monitoring  and Support
   Laboratory, U.  S. Environmental  Protection Agency,  Cincinnati, Ohio.

.Weltering,  D. M.   1984.   The growth  response in  fish  chronic and early
   life stage  toxicity tests:   A critical  review.   Aquat.  Toxicol.  5:1-21.

^Zaroogian,  G. E., G.  Pesch,  and G. Morrison.  1969.   Formulation of  an
   artificial  sea water  media suitable for oyster  larvae development.
   Amer.  Zool. 9:1141.

 Zillioux,  E.  J.,  H. R.  Foulk,  J.  C.  Prager,  and  J.  A.  Cardin.   1973.
   Using  Artemia  to assay oil  dispersant  toxicities.   JWPCF 45:2389-2396.
                                    318

-------
                                        APPENDIX


filA   Independence, Randomization and Outliers .	320

 |||B.   Validating Normality and Homogeneity of Variance
 ''*""    Assumptions	,	323

        1.  Introduction	323

        2.  Test for Normal  Distribution of Data	323

        3,  Test for Homogeneity of Variance	331

        4.  Transformations  of Data	332

  |C.   Dunnett's  Procedure	335

        1,  Manual  Calculations 	  335

        2.  Computer Calculations	342

ip D.   Bonferroni's T-test	  381
ffTI
I IE.   Steel's Many-one  Rank  Test	337-

U|F.   Wilcoxon Rank Sum Test	  .                 qq?
ji^|||s                                ••«••«••«««.**••...,,  •jjc.

        Probit  Analysis   .	393
                                        319

-------
                                   APPENDIX A

                   INDEPENDENCE, RANDOMIZATION, AND OUTLIERS?
 tl.  STATISTICAL INDEPENDENCE

 p.l  Dunnett's Procedure and Bonferroni's T-test are parametric procedures
 fbased on the assumptions that (1) the observations within treatments are
 [independent and normally distributed, and (2) that the variance of the
 observations is homogeneous across all toxicant concentrations and the
 tontrol.  Of the three possible departures from the assumptions,
 fnon-normality, heterogeneity of variance, and lack of independence, those
 paused by lack of independence are the most difficult to deal  with (see
 iScheffe, 1959).  For toxicity data, statistical independence means that
 fgiven knowledge of the true mean for a given concentration or control,
 fknowledge of the error in any one actual observation whould provide no
 ^informtion about the error in any other observation.  Lack of independence
 n's difficult to assess and difficult to test for statistically.  It may also
 Ihave serious effects on the true alpha or beta level.  Therefore,  it is of
 jutmost importance to be aware of the need for statistical independence
 |between observations and to be constantly vigilant in avoiding any patterned
 jexperimental procedure that might compromise independence.  One of the best
 [ways to help insure independence is to follow proper randomization
 procedures throughout the test.

 2.  RANDOMIZATION

 |2.1  Randomization of the distribution of test organisms among test vessels,
 [and the arrangement of treatments and replicate vessels is an  important part
 of conducting a valid test.  The purpose of randomization is to avoid
 situations where test organisms  are placed serially into test  vessels, or
 jWhere all replicates for a test  concentration are located adjacent to one
 .another, which could introduce bias into the test results.

 ;2.2  An example of randomization is described using the Sheepshead Minnow
 larval  Survival and Growth test.  For a test design with five  treatments,  a
[control, and three replicates at each treatment,  there would be 18
 ^experimental units,  i.e.,  18 positions to be randomized.   There are several
 sways to randomly assign the positions.  Random numbers may be  selected from
   random numbers table or  generated by computer software.

12.3  In this example,  the  first  three random numbers selected  would be used
por the three control  replicates.   The selction of random numbers  would be
^continued,  three at  a time, and  assigned to  a particular treatment,
^progressing  from the lowest to the highest test concentration.   The rank
[ordering of  these random numbers would determine  the relative  positioning
jfor the controls and concentration levels.
 ^Prepared by Ron Freyberg,  Florence  Kessler,  John  Menkedick and Larry
.Wymer,  Computer  Sciences  Corporation,  26  W. Martin Luther King Drive,
 Cincinnati,  Ohio 45268; Phone 513-569-7968.
                                      320

-------
2.4  The result of this randomization procedure  is presented in Table A.I,
using an effluent concentration series of  1.0%,  3.2%, 10.0%, 32.0%, and
100%.
 TABLE A.I.  RANDOMIZATION OF THE POSITIONS OF EXPERIMENTAL UNITS USING A
             DESIGN OF THREE ROWS AND SIX COLUMNS
|H 3.2%
BiQii
Hi 32.0%
Hf i>o%
mm 	
32.0%
10.0%
Control
3.2%
Control
10.0%
1.0%
3.2%
32.0%
100%
Control
100%
10.0%
1.0%
100%
3.  OUTLIERS

3.1  An outlier is an inconsistent or questionable data point that
appears unrepresentative of the general trend exhibited by the majority
of the data.  Outliers may be detected by tabulation of the data,
plotting, and by an analysis of the residuals.  An explanation should be
sought for any questionable data points.  Without an explanation, data
points should be discarded only with extreme caution.  If there is no
explanation, the analysis should be performed both with and without the
outlier, and the results of both analyses should be reported.

3.2  Gentleman Wilks1 A statistic gives a test for the condition that the
extreme observation may be considered an outlier.  For a discussion of
this, and other techniques for evaluating outliers, see Draper and John
(1981).
                                    321

-------
                       TABLE  A.2.    TABLE  OF  RANDOM  NUMBERS?
      09 73  25 33
   37  64 20  48 05
      42 26  89 63
      01 90  25 29
   IS  80 79  99 70

 |66  06 57  47 17
 f3I  06 01  08 05
  85  26 97  76 02
  63  57 33  21  35
  73  79 64  57  53

  98  52  01  77  67
  11  80  50 54  31
  83  45  29 96  34
  88  68  54 02  00
  99  59  46 73  48

 65 48  11  76  74
 80  12  43  56 35
 74 35  09  98  17
 69 91  62  68 03
 09 89  32  05 05

 91  49 91  45 23
 80 33 69 45 98
 44  10 48  19 49
 12 55 07 37  42
 63 60 64 93  29

 61 19 69 04  46
 15 47 44  52  66
 94 55  72  85  73
 42 48  11  £2  13
 23 52  37  83  17

 04 49  35  24 94
 00 54 99  76 54
 35 96 31  53 07
 59 80 80  83 91
 46 05 88  52 36

 32  17 90  05 97
 69  23 46  14 06
 19  56 54  14 30
45  15 51 49 38
W  86 43  19  94

98  08 62 48  26
33  18 51 62  32
80  95 10 04  06
79  75  24 91  40
18   63  33 25  37
   76 52 01  35 86
   64 89 47  42 96
   19 64 50  93 03
   09 37 67  07 15
   80 16 73  61 47

   34 07 27  68 50
   45 57 18  24 06
   02 05 16  56 92
  05 32 54  70 48
  O3 52 96  47 78

  14  90 56  86  07
  39  80 82  77  32
  06  28 89  80  83
  86  50 75 84  01
  87  51  76 49  69

  17  46  85 09  50
  17  72  70 80  15
  77  40  27  72  14
  66 25  22  91 48
  14 22  56  85  14

  68 47 92  76 86
  26 94 03 68 58
  85 15 74 79 54
  11  10 00 20 40
  16  50 53  44  84

 26  45  74  77  74
 95  27  07  99  53
 67  89  75  43  87
 97  34  40  87  21
 73  20  88  98  37

 75  24  63  38 24
 64 05  18  81 59
 26 89  80  93 54
 45 42  72  68 42
 01  39 09  22 86

 87  37 92  62 41
 20  11 74  52 04
 01  75 87  53 79
 19  47 60  72 46
36  16 81 08 51

 45  24 02 84  04
41  94 15 09  49
96  38 27 07  74
71  96 12 82  96
98  14 50 65  71
                                          34 67  35 48 76
                                          24 80  52 40 37
                                          23 20  90 25 60
                                          38 31  13 11 65
                                          64 03  23 66 53

                                          36 69  73  61 70
                                          35 30  34  26 14
                                          68 66  57  48 18
                                          90 55 35  75 48
                                          35 80 83  42 82

                                          22 10 94  05 68
                                          50 72 56 82  48
                                          13 74 67 00  78
                                          36 76 66 79  51
                                          91  82 60 89  28

                                          58  04 77 69  74
                                         45  31  82 23 74
                                         43  23  60 02  10
                                         36  93  68 72 03
                                         46  42  75 67 88

                                         46  16 28  35 54
                                         70 29 73  41  35
                                         32 97 92  65  75
                                         12 86 07 46  97
                                         40 21  95 25  63

                                         51  92  43 37  29
                                         59  36  78 38  48
                                         54  62  24 44  31
                                         16  86  84 87 67
                                        68  93  59 14  16
                                       80 95  90 91 17
                                       20 63  61 04 02
                                       15 95  33 47 64
                                       88 67  67 43 97
                                       ?S 95  11 88 77

                                       65 81  33 98 85
                                       86 79  90 74 39
                                       73 05  38 52 47
                                       28 46 82  87 09
                                       60 93 52  03 44

                                       60 97 09 34  33
                                       29 40 52 42  01
                                       18 47 54 06  10
                                       90 36 47 64  93
                                       93 78 56 13  68

                                       73  03 95 71  86
                                      21  II  57 82 53
                                      45  52  16 42 37
                                      76 62  11 39 90
                                      96 29 77 88 22

                                      94 75 08  99 23
                                      53  14  03  33  40
                                      57  60  04  08  $1
                                      96  64  48  94  39
                                      43  65  17  70  82

                                      65  39  45  95  93
                                      82  39  61 01  18
                                      91  19  04 25  92
                                     03  0?  II 20  59
                                     26  25  22 96 63
                                        45 86 25  10 25   61  96 27  93  35
                                        96 II 96  38 96   54  69 28  23  91
                                        33 35 13  54 62   77  97 45  00  24
                                        83 60 94  97 00   13  02 12  48  92
                                        77 28 14  40 77   93  91 08  36  47
                                        05 56 70  70  07
                                        15 95 66  00  00
                                        40 41 92  15  85
                                        43 66 79  45  43
                                        34 88 88  15  53

                                        44 99 90  88  96
                                        69 43  £4 85  81
                                        20 15 12 33  87
                                        69 86  10 25  91
                                        31  01  02 46  74
                                     86 74 31 71  57
                                     18 74 39 24  23
                                     66 67 43 68  06
                                     59 04 79 00  33
                                     01  54 03 54  56
                                       39 29 27 49  45
                                       00 82 29 16  65
                                       35 08 03 36  06
                                       04 43 62 76  59
                                       12 17 17 68  33

                                       11  19 92 91  70
                                       23  40 30 97  32
                                       18  62 38 85  79
                                       83  49 12 56 24
                                       35  27 38 84 35

                                      50  50 0? 39 98
                                      52  77  56 78 51
                                      68  71  17 78 17
                                      29 60  91  10 62
                                      23 47  83  41  13

                                      40 21  81  65 44
                                      14 38  55  37 63
                                      96 28 60  26 55
                                      94 40 05 64 18
                                      54 38 21  45 98

                                      37 08 92  00  48
                                      42  05 08  23  41
                                     22 22 20  64  13
                                     28 70 72  58  15
                                     07 20  73  17  90

                                     42 58  26  05  27
                                     33 21  !5  94  66
                                     92 92  74  59  73
                                     25 70  14  66  70
                                     05 52 28  25  62

                                     65 33 71 24 72
                                     23 28 72 95 29
                                     90 10 33 93 33
                                     78 56 52 01 06
                                     70 61 74 29 41

                                     85  39 41 18 38
                                     97  11  89 63 38
                                     84  96 2& 52 07
                                     20  82  66 95 41
                                     05  01  45 11 76
                                     39  09 47 34 07   35 44  13  18  80
                                     88  69 54 19 94   37 54  87  30  43
                                     25  01 62 62 98   94 62  46  11  71
                                     74  85 22 05 39   00 38  75  95  7fl
                                     05  45 56 H 27   77 93  89  19  38
   74 02 94  39 02
   64 17 84  56 11
   11 66 44  98 83
   48 32 47  79 28
   69 07 49  41  38
77  55 73 22  70
80  99 33 71  43
52  07 98 48  27
31  24 96 47  10
87  63 79 19  76
97 79  01  71 19
05 33  51  29 69
59 38  17  15 39
02 29  53  68 70
35 68  40  44 01
                                                      52 52  75  80 21
                                                      56 12  71  92 £5
                                                      09 97  33  34 4O
                                                      32 30  75  75 46
                                                      10 51  82  16 15
                                                      80 81  45 17 48
                                                      36 04  09 03 24
                                                      88 46  12 33 56
                                                      15 02  00 99 94
                                                      01  Si  87 69 38
^From Dixon  and  Massey,   1983.
                                          322

-------
                                   APPENDIX  B

          VALIDATING NORMALITY AND HOMOGENEITY OF VARIANCE ASSUMPTIONS!
     INTRODUCTION

  .1   Dunnett's  Procedure  and  Bonferroni's T-test are parametric procedures
 Jased on the assumptions  that the observations within  treatments  are
 independent and normally  distributed,  and that the variance of the
 [bservations is homogeneous across  all toxicant concentrations and the
 Sontrol.  These assumptions should  be  checked prior to using these tests, to
 fetermine if they  have  been met.  Tests for  validating the assumptions  are
 rovided in the following discussion.  If the tests fail  (if the  data do not
 eet the assumptions),  a  non-parametric procedure such as Steel's Many-One
jlank Test may be more appropriate.   However, the decision on whether to use
 larametric or non-parametric  tests  may be a  judgement  call, and a statistician
[should  be consulted  in  selecting the analysis.

     TEST FOR NORMAL  DISTRIBUTION OF DATA

[2.1   A  formal test for  normality is the Shapiro-Wilks  Test (1).   The test
^statistic is obtained by  dividing the  square of an appropriate linear
|combination of  the sample order statistics by the usual symmetric estimate of
[variance.  The  calculated W must be greater  than zero  and less than or  equal
[to one.  This test is recommended for  a sample size of 50 or less.  If  the
tsample  size is  greater  than 50, the Kolomogorov "D" statistic  (2) is
[recommended.  An example  of the Shapiro-Wilks test is  provided below.

[2.2   The example uses growth  data from the Sheepshead  Minnow Larval Survival
jand  Growth Test.   The same data are used  in  the discussion of the homogeneity
[of variance determination in  Paragraph 3  and Dunnett's Procedure  in
^Appendix C.  The data and the mean  and standard deviation of the  observations
|at each concentration,  including the control, are listed  in Table B.I.

'2.3   The first  step  of  the test for normality is to center the observations by
'subtracting the mean from each observation at its respective concen- tration
[and  control.  The  centered observations are  listed in  Table B.2.

[2.4   Calculate  the denominator, D,  of  the test statistic:
                                 „
                      D  =   L  (X.- X
    Where:  X-j  =     the  centered  observations  and  X  is  the overall mean^ of
                    the  centered  observations.   For  this  set of data,  X = 0,
                    and  D  =  0.1589.
^Prepared  by  Ron  Freyberg,  Florence  Kessler,  John Menkedick and Larry
'Wymer,  Computer Sciences  Corporation,  26  W. Martin Luther King Drive,
'Cincinnati, Ohio  45268; Phone  513-569-7968.

                                      323

-------
 2.5  Order the centered observations from smallest  to  largest
                                                     (n)
     and  let  X-j  denote  the  1th  order  statistic.   The ordered observations
 are  listed  in Table  B.3.

 2,6   From Table B.4, for the number  of  observations, n, obtain the
^coefficients a^  a2, 	,  ak, where k  is approximately n/2.  For the

 data  in  this example,  n =  15,  k = 7,  and the a*  values are listed in
 Table B.5.
 2.7   Compute the test statistic, W, as follows:
                             1 = 1
                                 a.
                                  7
    The differences, x(n-i+l) - xH), are listed in Table B.5.

2.8  The decision rule for this test is to compare the critical value from
Table B.6 to the computed W.  If the computed value is less than the
critical value, conclude that the data are not normally distributed.  For
this example, the critical value is 0.835.  The calculated value, 0.9516, is
not less than the critical value.  Thus, the conclusion of the test is that
the data are normally distributed.

2.9  In general, if the data fail the test for normality, a transformation
such as to log values may normalize the data.  After transforming the data,
repeat the Shapiro-Wilks Test for normality.
                                 \
324

-------
       TABLE  B.I.   SHEEPSHEAD  LARVAL GROWTH  DATA  {WEIGHT IN MG)
                    FOR  THE  SHAPIRO-WILKS TEST
H Effluent Concentration (%)
H
• Replicate
ii i
H 2
11;
II 3
11
ii Mean
Hi

If *
Control
1.017
0.745

0.862
0.875
0.14
1
1.0
1.157
0.914

0.992
1.021
0.12
2
3.2
0.998
0.793

1.021
0.937
0.13
3
10.0
0.837
0.935

0.839
0.882
0.05
4
32.0
0.715
0.907

1.044
0.889
0.17
5
TABLE B.2.  EXAMPLE OF SHAPIRO-WILKS TEST CENTERED OBSERVATIONS
m
&m
|p
KReplicate Control
I
| 1 0.142
9.
m 2 ~ 0<13
1 3 " °*013

1.0
0.136
- 0.107
- 0.029
Effluent Concentration
3.2 10.0
0.061 - 0.009
- 0.144 0.053
0.084 - 0.043
{*>
32.0
- 0.174
0,018
0.155

                                   325

-------
TABLE B.3. EXAMPLE OF THE SHAPIRO-WILKS TEST: ORDERED OBSERVATIONS
                                              xd>
                1
                2
                3
                4
                5
                6
                7
                8
                9
                10
                n
                12
                13
                14
                15
0.174
0.144
0.130
0.107
0.043
0.029
0.013
0.009
0.018
0.053
0.061
0.084
0.136
0.142
0.155
                                  326

-------
TABLE B.4.  COEFFICIENTS FOR   THE  SHAPIRO-WILKS
 0.7071
0.7071
0.0000
0.6872
0.1667
0.6646
0.2413
0.0000
0.6431
0.2806
0.0875
0-6233
0.3031
0.1401
0.0000
0 6052
03164
0.1743
00561
0 5888
03244
0 1976
0 0947
0 0000
0 5739
03291
02141
0 1224
    11
           12
                          14
                                  15
                                         16
                                                17
                                                               19
                                                              20
1
2
3
4
5
6
7
8
9
10
0.5601
0.3315
0.2260
0.1429
0.0695
0.0000
0.5475
0.3325
0.2347
0.1586
0.0922
0.0303
0.5359
0.3325
0.2412
0.1707
0 1099
0.0539
0.0000
0.5251
0.3318
0.2460
0.1802
0.1240
0.0727
0.0240
0.5150
0.3306
0.2495
0.1878
0.1353
0.0880
0.0433
0.0000
0.5056
0,3290
0.2521
0,1939
0.1447
0.1005
0.0593
0.0196
04968
0.3273
0.2540
0.1988
0.1524
0.1109
0.0725
0.0359
0.0000
04886
0.3253
02553
0.2027
0 1587
0 1197
0 0837
0.04%
OOV63
0.4808
0.3232
0,2561
0 2059
0.1641
0 1271
0 0932
0.0612
ft 0303
0 0000
0.4734
0.3211
0.2565
0.2085
0.1686
0.1334
0.1013
0.0711
0.0422
0.0140
    0.4643
    0.3185
    0.2578
    0.2119
    0.1736
    0,1399
    0.1092
 8  0.0804
 9  0,0530
 10  0.0263
 11  0.0000
 12    —
 13    —
 14   —
 IS   —
   0.4590
   0.3156
   0.2571
   0.2131
   0.1764
   0.1443
   0.1150
   0.0878
   0.0618
   0-0368
   0.0122
   0.4542
   0.3126
   0.2563
   0.2139
   0.1787
   0.1480
   0.1201
   0.0941
   0.0696
   0.0459
   0.0228
   0.0000
   0.4493
   0,3098
   0.2554
   0.2145
   0.1807
   0.1512
   0.1245
   0.0997
   0.0764
   0.0539
   0.0321
   0.0107
   0.4450
   0.3069
   0.2543
   0.2148
   0.1822
   0.1539
   0.1283
   0,1046
   0.0823
   0.0610
   0.0403
   0.0200
   0.0000
   0.4407
   0.3043
   0.2533
   0.2151
   0.1836
   0.1563
   0.1316
   0.1089
   0.0876
   0.0672
   0.0476
   0,0284
   0.0094
   0.4366
   0.3018
   0.2522
   0.2152
   0.1848
   0.1584
   0.1346
   0.1128
   0.0923
   0.0728
   0.0540
   0,0358
   0.0178
   0.0000
   0.4328
   0.2992
   0.2510
   02151
   0.1857
   0.1601
   0.13^2
   0.1162
   0.0965
   0.0778
   0.0598
   0.0424
   0.0253
   0.0084
   0.4291
   0.2968
   0.2499
   0.2150
   0.1864
   0.1616
   0.1395
   0,1192
   0.1002
   0.0822
   0.0650
   0.0483
   0.0320
   0.0159
   0.0000
0.4254
0.2944
0.2487
0.2148
0.1870
0.1630
0.1415
0.1219
0.1036
0.0862
0.0697
0.0537
 0.0381
 0.0227
 0.0076
            iTaken  from:  Conover,  1980,
                                      327

-------
TABLE B.4  COEFFICIENTS FOR THE SHAPIRO-WILKS  TEST  (Continued)
V
\
1 N
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20

'\n

1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25

31
\
0.4220
0.2921
0.2475
0.2145
0 1874
0.1641
0.1433
0 1243
0.1066
0.0899
0.0739
0.0585
0 0435
0.0289
0.0144
0.0000
—
__
—
—


41
\
0.3940
0.2719
0.2357
0.2091
0.1876
0.1693
0.1531
0.1384
0.1249
0.1123
0.1004
0.0891
0.0782
0.0677
0.0575
0.0476
0.0379
0.0283
0.0188
0.0094
0.0000

—
—
—

32

0.4188
0.2898
0.2462
0.2141
0.1878
0.1651
0.1449
0.1265
0.1093
0.0931
0.0777
0.0629
0.0485
0.0344
0.0206
0.0068
—
—
_.
—


42

0.3917
0.2701
0.2345
0.2085
0.1874
0.1694
0.1535
0.1392
0.1259
0.1136
0.1020
0.0909
0.0804
0.0701
0.0602
0.0506
0.0411
0.0318
0.0227
0.0136
0.0045
—
—
—
—

33

0.4156.
0.2876
0.2451
0.2137
0.1880
0.1660
0.1463
0.1284
0.1118
0.0961
0.0812
0.0669
0,0530
0.0395
0.0262
0.0131
0.0000
—
—
—


43

0.3894
0.2684
0.2334
0.2078
0.1871
0.1695
0.1539
0.1398
0.1269
0.1149
0.1035
0.0927
0.0824
0.0724
0.0628
0.0534
0.0442
0.0352
0.0263
0.0175
0.0087
0.0000
—
—
—

34

0.4127
0.2854
0.2439
0.2132
0.1882
0.1667
0.1475
0.1301
0.1140
0.0988
0.0844
0.0706
0.0572
0.0441
0.0314
0.0187
0.0062
—
—
—


44

0.3872
0"2667
0.2323
0.2072
0.1868
0.1695
0.1542
0.1405
0.1278
0.1160
0.1049
0.0943
0.0842
0.0745
0.0651
0.0560
0.0471
0.0383
0.0296
0.0211
0.0126
0.0042
—
—
—

35

0.4096
0.2834
0.2427
0.2127
0.1883
0.1673
0,1487
0.1317
0.1160
0.1013
0.0873
0.0739
0.0610
0.0484
0.0361
0.0239
0.0119
0.0000
—
—


45

0-3850
0.2651
0.2313
0.2065
0.1865
0.1695
0.1545
0.1410
0.1286
0.1170
0.1062
0.0959
0.0860
0.0765
0.0673
0.0584
0.0497
0.0412
0.0328
0.0245
0.0163
0.0081
0.0000
—
—

36

0.4068
0.2813
0.2415
0.2121
0.1883
0.1678
0.1496
0.1331
0.1179
0.1036
0.0900
0.0770
0.0645
0.0523
0.0404
0.0287
0.0172
0.0057
„
—


46

0.3830
0.2635
0.2302
0.2058
0.1862
0.1695
0,1548
0.1415
0.1293
0.1180
0.1073
0.0972
0.0876
0.0783
0.0694
0.0607
0.0522
0.0439
0.0357
0.0277
0.0197
0.0118
0.0039
—
—

37

0.4040
0.2794
0.2403
0.2116
0.1883
0.1683
0.1505
0.1344
0.1196
0.1056
0.0924
0.0798
0.0677
0.0559
0.0444
0.0331
0.0220
0.0110
0.0000
—


47

0.3808
0.2620
0.2291
0.2052
0.1859
0.1695
0.1550
0.1420
0.1300
0.1189
0.1085
0.0986
0.0892
0.0801
0.0713
0.0628
0.0546
0.0465
0.0385
0.0307
0.0229
0.0153
0.0076
0.0000
—

38

0.4015
0.2774
0.2391
0.2110
0.1881
0 1686
0.1513
0.1356
0.1211
0.1075
0.0947
0.0824
0.0706
0.0592
0.0481
0.0372
0.0264
0.0158
0.0053
—


4$

0.3789
0.2604
0.2281
0.2045
0.1855
0.1693
0 t5M
0.1423
0.1306
0.1197
0.1095
0.0998
0.0906
00817
0.0731
0.0648
0 0568
.0.0489
0.0411
0.0335
0.0259
0.0185
0.0111
0.0037
—

39

0.3989
0.2755
0.2380
0.2104
0.1880
0.1689
0.1520
0.1366
0.1225
0.1092
0.0967
0.0848
0.0733
0.0622
0.0515
0.0409
0.0305
0.0203
0.0101
0.0000


49

0.3770
0.2589
0.2271
0.2038
0.1851
0.1692
0.1553
0.1427
0.1312
0.1205
0.1105
0.1010
0.0919
0.0832
0.0748
0.0667
0.0588
0.05 3 1
0.0436
0.0361
0.0288
0.0215
0.0143
0.0071
0.0000

40

0.3964
0.2737
0.2368
0.2098
0.1878 ;•'
0.1691 ;
01526
0.1376
0.1237
0.1108
0.0986
0.0870
0.0759
0.0651
0.0546
0.0444
0.0343
0.0244
0.0146
0.0049


50

0.3751
0.2574
0.2260
0.2032
0.1847
0.1691
0.1554
0.1430
0.1317
0.1212
0.1113
0.1020
0.0932
0.0846 .
0.0764 .-.'
0.0685
0.0608
0.0532
0.0459
0.0386
0.0314
0.0244
0.0174
0.0104
0.0035
                           328

-------
TABLE B.5. EXAMPLE OF THE SHAPIRO-WILKS TEST:
           TABLE OF COEFFICIENTS AND DIFFERENCES
             0.5150
             0.3306
             0.2495
             0.1878
             0.1353
             0.0880
             0.0433
0.329
0.286
0.266
0.191
0.104
0.082
0.031
                     329

-------
                                                 ^
TABLE B.6 QUANTILES OF THE SHAPIRO-W1LKS TEST  STAT&IC1
n
3
4
5
6
7
8
9
10
11
12
13
14
IS
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
0.0 1
0.753
0.687
0.686
0.713
0.730
0.749
0.764
0.781
0.792
0.805
0.814
0.825
0.835
0.844
0.851
0.858
0.863
0.868
0.873
0.878
0.881
0.884
0.888 •
0.891
0.894
0.896
0.898
0.900
0.902
0.904
0.906
0.908
0.910
0.912
0.914
0.916
0.917
0.919
0.920
0.922
0.923
0.924
0.926
0.927
0.928
0.02
0.756
0.707
0.715
0.743
0.760
0.778
0.791
0.806
0.817
0.828
0.837
0.846
0.855
0.863
0.869
0.874
0.879
0.884
0.888
0.892
0.895
0.898
0.901
0.904
0.906
0.908
0.910
0.912
0.914
0.915
0.917
0.919
0.920
0.922
0.924
0.925
0.927
0.928
0.929
0.930
0.932
0.933
0.934
0.935
0.936
0.05
0.767
0.748
0.762
0.788
0.803 .
0.818
0.829
0.842
0,850
0.859
0.866
0.874'
0.881
0.887
0.892
0.897
0.901
0.905
0.908
0.911
0.914
0.916
0.918
0.920
0.923
0.924
0.926
0.927
0.929
0.930
0.931
0,933
0.934
0.935
0.936
0.938
0.939
0.940
0.941
0.942
0.943
0.944
0.945
0.945
0.946
O.JO
0.789
0.792
0.806
0.826
0.838
0.851
0.859
0.869
0.876
0.883
0.889
0.895
0.90.1
0.906
0.910
0.914
0.917
0.920
0.923
0.926
0.928
0.930
0.931
0.933
0.935
0.936
0.937
0.939
0.940
0.941
0.942
0.943
0.944
0.945
0.946
0.947
0.948
0.949
0.950
0.951
0.951
0.952
0.953
0.953
0.954
0.50
0.959
0.935
0.927
0.927
0.928
0.932
0.935
0.938
0.940
0.943
0.945
0.947
0.950
0.952
0.954
0.956
0.957
0.059
0.960
0.961
0.962
0.963
0.964
0.965
0.965
0.966
0,966
0.967
0.967
0.968
0.968
0.969
0.969
0.970
0.970
0.971
0.971
0.972
0.972
0.972
0,973
0,973
0.973
0.974
0.974
0.90
0.998
0.987
0.979
0.974
0.972
0.972
0.972
0.972
0.973
0.973
0.974
0.975
0.975
0.976
0.977
- 0.978
0.978
0.979
0.980
0.980
0.981
0.981
0.981
0.982
0.982
0,982
0.982
0.983
0.983
0.983
0.983
0.983
0.984
0.984
0.984
0.984
0.984
0.985
0.985
0.985
0.985
0.985
0.985
0.985
0.985
0.95
0.999
0.992
0.986
0.981
0.979
0.978
0.978
0.97H
0.979
0.979
0.979
0.980
0,980
0.981
0.981
0.982
0.982
0.983
0.983
0.984
0.984
0.984
0.985
0.9&5
0.9&5
0.985
0.985
0.985
0.986
0.986
0986
0.986
0.986
0.986
0.98^
0.98"*
0.98"
0.98"
0.98"
098"
098"
0.98"
0.988
0.988
0.98S
0.98 -
1.000
0.996
0.991
0.986
0.985
0.984
0.984
0.983
0.984
0.984
0.984
0.984
0.984
0.985
0.985
0.986
0.986
0.986
0.987
0.987
0.987
0.987
0.988
0.988
0.988
0.988
0.988
0.988
0.988
0.988
0.989
0.989
0.989
0.989
0.989
0.989
0.989
0.989
0.989
0.989
0.990
0.990
0.990
0.990
0.990
a|8
*S8$L-
lOOt
•-SSlS;1
0.9®
tfi-^7**
0.99,3%
0.98||
0.98ft
0.987^
0.986f
0.986 t
0.986
0.986
0.986
0.986
0.987
0.987
0.987
0.988
0.988
0.988
0.989
0.989
0.989
0.989
0.989
0.989
0.990
0,990
0.990
0.990
0.990
0.990
0.990
0.990
0.990
0.990
0.990
0.990
0.991
0.991
0.991
0.991
0.991
0.991
0.991
0.991
0.991
      iTaken from Conover, 7980.
                          330

-------
[3.   TEST FOR HOMOGENEITY OF VARIANCE
£                                                            '•' :'
;3.1   For Dunnett's Procedure and Bonferroni's  T-test,  the  variances  of  the
data obtained from each  toxicant concentration and  the control  are assumed
    be equal.  Bartlett's Test is a formal  test of .this assumption.   In  using
this test,  it is  assumed that the data  are normally distributed.
i

,3.2   The data used in  this  example are  growth  data  from a  Sheepshead Minnow
|Larval  Survival and Growth  Test, and  are the same data used in  Appendices
•C  and D. These data are listed  in Table B.7,  together with the calculated
standard deviation for the  control and  each toxicant concentration.

;3.3   The test statistic  for Bartlett's  Test (Snedecor  and  Cochran, 1980) is
'as follows:
            B  =
                Z V, In S }
                i
                               I  V.  In  S.
                               1
Where: V
       C
       In
             =  Degrees  of  freedom for  each  toxicant concentration and control
             =  Number of  levels of  toxicant concentration  including the
               control

             =  The average of  the individual variances.

             =  1 + [l/3(a-l)][S 1/V1 -  1/Z  V.]
                              "
             = Loge

3.4  Since B is approximately distributed as chi-square with a - 1 degrees
of freedom when the variances are equal, the appropriate critical value  is
obtained from a table of the chi-square distribution for a - 1 degrees of
freedom and  a significance  level of 0.01.  If B is less than the critical
value then the variances are assumed to be equal.
                                                  _2
3.5  For the data in this example, v-j = 2, a = 5, S  = 0.0158, and
C =  1.2.  The calculated B  value is:
                 B =
                     2 [5{ln 0.0158) - I In Si
                                 1.2

                     2[5(- 4.1477) - (- 22.1247)]
                               1.2
                   = 2.3103
                                     331

-------
  I
  I 3.5  Since  B  is approximately distributed as chi-square with a - 1 degrees
    ?  f™edoni  when the variances are equal, the appropriate critical value
    tor the test  is 13.3 for a significance  level of 0.01.  Since B  < 13 3
    the conclusion is that the variances are equal.                     * *
        TABLE B.7.  SHEEPSHEAD LARVAL GROWTH DATA (WEIGHT IN MG) USED FOR
                    BARTLETT'S TEST FOR HOMOGENEITY OF VARIANCE
    Replicate
Control
Effluent Concentration (%)
      37F10.0      32"
1 1 1.017
1 2 0.745
i
I 3 0.862
i
I Mean 0.875
i
1 ^ ' 0.14
i i
§
1.157
0.914
0.992
1.021
0.12
2
0.998
0.793
1.021
0.937
0.13
3
0.837
0.935
0.839
0,882
0,05
4
0.715
0.907
1.044
0.889
0.17
5
|| 4.   TRANSFORMATIONS OF THE DATA
f-^SJ.^
fl-^i^3~^
|| 4.1  When the assumptions of normality and/or homogeneity of variance are
Ijnot  met, transformations of the data may remedy the problem, so that the
61 ata can be ana^yzed by parametric procedures, rather than a
P| non-parametric technique such as Steel's Many-one Rank Test or Wilcoxon's
BjKank Sum Test.  Examples of transformations include log, square root  arc
|| .sine .square root, and reciprocals.  After the data have been transformed,
||Shapiro-Wilks and Bartlett's tests should be performed on the transformed
m observations to determine whether the assumptions of normality and/or
M homogeneity of variance are met.
^^3'-
04.2  Arc Sine Square Root Transformation^
   4.2.1  For data consisting of proportions from a binomial  (response/no
   response; live/dead) response variable, the variance within the i-th
   treatment is proportional to P^ (1 - PJ), where P,  is the  expected
   proportion for the treatment.  This clearly violates the homogeneity of
 I _ , __ :
   ^rom:  Peltier and Weber (1985).
                                       332

-------
 variance assumption required by parametric procedures such as Dunnett's or
 Bonferroni's, since the existence of a treatment effect Implies different
 values of Pi for different treatments, i.  Also, when the observed
 proportions are based on small samples, or when Pi is close to zero or
 one, the normality assumption may be invalid.  The arc sine square root
 Urcsin - P) transformation is commonly used for such data to stabilize the
 variance and satisfy the normality requirement.

 4.2.2  Arc sine transformation consists of determining the angle  (in
 radons)  represented by a sine value.   In the case of arc sine square  root
 transformation of mortality data,  the  proportion of dead  (or  affected)
 organisms is taken as the sine value,  the square root of  the  sine  value is
 determined,  and the angle (in radians)  for the square root of the  sine  value
 is  determined.   Whenever the proportion dead is 0 or 1, a special
 modifica^on of the arc  sine square  root  transformation must  be used

                                          3rC Slne '<"»'••  ™*  transformation
4.2,3  Calculate  the response proportion  (RP) at each effluent
concentration, where:

    RP =  (number  of dead or "affected" organisms)/ (number exposed),

    Example:  If  8 of 20 animals in a given treatment die:

              RP  = 8/20
                  = 0.40

4.2.4  Transform  each RP to arc sine,  as follows.

4.2.4.1  For RPs greater than zero or  less than one:

         Angle (radians) = arc sine (RPJ0.5.

         Example:  If RP = 0.40:                                /

                  Angle = arc  sine  (0.40)0-5     ^/[    /):

                        = arc  sine  0.6325

                        = 0.6847  radians
                                     333

-------
4.2.4.2  Modification of the arc sine when RP = 0.



         Angle  (in radians) = arc sine (1/4NJ0.5



         Where: N = Number of animals/treatment



         Example: If 20 animals are used:



                  Angle = arc sine (1/80)°«5



                        = arc sine 0.1118



                        = 0.1120 radians





4.2.4,3  Modification of the arc sine when RP = 1.0.



         Angle - 1.5708 radians - (radians for RP  = 0)



         Example:  Using above value:



                  Angle = 1.5708 -  0.1120



                        = 1.4588 radians
                                     334

-------
                                  APPENDIX C

                              DUNNETT'S PROCEDURE
li.  MANUAL CALCULATIONS!

 :;.l  Dunnett's Procedure  is used to compare each concentration mean with the
 ontrol mean to decide if any of the concentrations differ from the
 Control.  This test has an overall error rate of alpha, which accounts for
    multiple comparisons with the control.  It is based on the assumptions
  lat the observations are independent and normally distributed and that the
  iriance of the observations is homogeneous across all concentrations and
  introl.  (See Appendix B for a discussion on validating the assumptions).
 Stinnett's Procedure uses a pooled estimate of the variance, which is equal
 |p the error value calculated in an analysis of variance.  Dunnett's
  •ocedure can only be used when the same number of replicate test vessels
 fave been used at each concentration and the control.  When this condition
   not met, Bonferroni's T-test is used (see Appendix D).

 12  The data used in this example are growth data from a Sheepshead Minnow
 arval Survival and Growth Test, and are the same data used in Appendices B
:aiid D.  These data are listed in Table C.I.  One way to obtain an estimate
 jf the pooled variance is to construct an ANOVA table including all sums of
 quares, using the following formulas:
        TABLE C.I. SHEEPSHEAD LARVAL GROWTH DATA (WEIGHT IN MG)
                   USED FOR DUNNETT'S PROCEDURE
1
Effluent
Cone (%)
Control
1.0
3.2
• 10.0
32.0
i
1
2
3
4
5
Replicate
1
1.017
1.157
0.998
0.873
0.715
Test
2
0.745
0.914
0.793
0.935
0.907
Vessel
3
0.862
0.992
1.021
0.839
1.044 .
Total
TT
2.624
3.063
2.812
2.647
2.666
Mean
Yi
0.875
1.021
0.937
0.882
0.889
^Prepared by Ron Freyberg, Florence Kessler, John Menkedick and Larry
Wymer, Computer Sciences Corporation, 26 W. Martin Luther King Drive,
Cincinnati, Ohio 45268; Phone 513-569-7968.

                                    335

-------
 1.3  One way to obtain an estimate  of  the pooled variance is to construct an
 ANOVA table including all sums of squares, using the following formulas:
 Total Sum of Squares:   SST = z  Y?. -  G2/N
                              ij  1J
 Between Sum of Squares: SSB = Z T?/n.  -  G2/N
                              j.i  i
 Within Sum of Squares:  SSW = SST - SSB
    Where: G = The grand total of all sample observations; G = I T.
           N = The total sample size; N = I n.                 1
          n.j = The number of replicates for concentration "i".
          TJ = The total of the replicate measurements for concentration ''i".
         Y.-.J = The jth observation for concentration "i".
          • j
1.4  Calculations:
Total  Sum of Squares:    SST = l  Y2.  -  G2/N
                                        13.812'
                             = 12.922 -
                             = 0.204
letween Sum of Squares:   SSB = I  T?/n. -  G2/N
|fithin Sum of Squares:
                             = 12.763 - (13.812)/15
                             = 0.045
                         SSW = SST - SSB
                             = 0.204 - 0.045
                             = 0.159
                                       336

-------
  5  Prepare the ANQVA table as follows:
                      TABLE C.2  GENERALIZED ANOVA TABLE
Bliource DF Sum of
B| Squares (SS)
if
IB: *
^Between b - 1 SSB
SSm;
fflBI-
•iithin N - b SSW
HmfP
Mean Square (MS)
(SS/DF)
Sp = SSB/(b-l)
D
$5 = SSW/{N-b)
w
Bfotal N - 1 SST
   = Number of different concentrations,  including  the  control
1.6  The completed ANOVA table for this data is provided below;
     TABLE C.3.  COMPLETED ANOVA TABLE FOR DUNNETT'S PROCEDURE
• " —
•source DF SS
•
jHiJjBi
IfBetween 5-1=4 0.045
• Within 15 - 5 = 10 0.159
mi
W 	 	
Mean Square
0.011
0.016
Total
14
0.204
                                      337

-------
jl.7  To  perform  the  individual comparisons, calculate the t statistic for
[each concentration and control combination, as follows:
                               - ,.
                                         +  (1/n.)
    Where: Yj    =  Mean  for each concentration

           Y]    =  Mean  for the control

           Sw    ~  Square root of the within mean square

           n-|    =  Number of  replicates in the control.

           n-j    =  Number of  replicates for concentration "i".

P.8  Table C.4   includes the  calculated t values for each concentration and
 ;ontrol combination.


                       TABLE  C.4.  CALCULATED T VALUES..
           Effluent
           Concentration

1.0
1 3*2
I 10.0
m
1 32.0
2
I 3
4
5
- 1.414
- 0.600
- 0.068
- 0.136
                                      338

-------
 1.9  Since the purpose of the test is only to detect a decrease in growth
 !from the control, a one-sided test is appropriate.  The critical value for
 (the one-sided comparison (2.47), with an overall alpha level of 0.05,
 HO degrees of freedom and four concentrations excluding the control is read
 rfrom the table of Dunnett's "T" values (Table D.5; this table assumes an
 Jequal number of replicates in all treatment concentrations and the
 [control).  Comparing each of the calculated t values in Table C.4 with the
 Ibritical value, no decreases in growth from the control were detected.  Thus
 She NOEC is 32.0%.

  .10  To quantify the sensitivity of the test, the minimum significant
  lifference (MSD) may be calculated.  The formula is as follows:
                           MSD = d Sw7(l/n1) + (l/n)
     Where: d   = Critical value for the Dunnett's Procedure

            Sw  = The square root of the within mean square

            n   = The number of replicates at each concentration,
                    assuming an equal  number of replicates  at  all
                    treatment concentrations

            n-j   = Number of replicates in the control

     For example:
         MSD = 2.47 (0.126)  7(1/3)  + (1/3)   =  2.47  (0.126)  72/3

             = 2.47 (0.126H0.816)

             = 0.254

|1.11   For this set of  data,  the  minimum  difference  between the control mean
|and a concentration mean  that  can be detected  as  statistically significant
  s  0.254 mg.   This represents  a  decrease in  growth  of 29% from the control.

  .11.1   If the data have  not been transformed, the  MSD  (and the percent
Decrease from the  control mean that  it represents)  can  be reported as is.

  .11.2   In the case where the  data have  been transformed, the MSD would be
In  transformed units.   In this case  carry out  the following conversion to
[determine the MSD  in untransformed units.
                                      339

-------
jl.11.2.1   Subtract the MSO from the transformed  control  mean.   Call  this
^difference 0.   Next,  obtain untransformed values for  the control  mean  and
iithe difference,  D.
 /here:
            MSDU  =  Controlu - Du



          MSDU = The minimum significant difference for untransformed data

      Controlu = The untransformed control  mean

            Du = The untransformed difference

|.n.2.2  Calculate the percent reduction from the control  that MSDU
represents as:
                               MSD,
          Percent  Reduction  =
                                           X  100
                               Controlu
11.11.3   An  example  of a conversion  of  the MSD to untransformed units, when
|the arcsin  square root transformation  was used on the data, follows.

    Step 1.  Subtract the MSD  from the  transformed control mean.  As an
             example, assume the data in Table C.I were transformed by the
             arc  sine square root transformation.  Thus:

                            0.875 - 0.254 = 0.621

    Step 2.  Obtain  untransformed values for the control mean (0.875) and the
             difference {0.621} obtained in Step 1, above.

                 [Sin(0.875)]2  =  0.589
                 [Sin(0.621)]2  =  0.339

    Step 3.  The  untransformed MSO (MSDU) is determined by subtracting the
             untransformed values obtained in Step 2.

             MSDU  =  0.589 - 0.339  =  0.250

       In this case, the MSD would represent a 42% decrease in survival  from
       the control  [(0.250/0.589H100)].
                                      340

-------
•1.12    Table of Dunnett's "t" values.
                        TABLE C.5.  DUNNETT'S "T" VALUESl
                                      (One-tailed) d
$$&K
•5k
mrap
HSU?
MHliiDz
HHH9&
Hi8
H|p
HE;10
m&&n
v&m™
HK?3
ffiHp4
«HP5
iffll16
IB"
IHh8
m&jLw
H20
mm2*
r^K30
mm, 40
1^60
fflE120
ire^Bf
B?£ 	

t
2.02
1.94
1.89
1.86
1.83
1.81
1.80
1.78
1.77
1.76
1.75
1.15
1.74
1.73
1.73
1.72
1.7J
1.70
1.68
1.67
1.66
1.64'

2
2.44
2.34
2.27
2.22
2.18
2.15
2.13
2.11
2.09
2.08
2.07
2.06
2.05
2.04
2.03
2.03
2.01
1.99
1.91
1.85
1.93
1.92

3
2.68
2.56
2.46
2.42
2.37
2.34
2.31
2.29
2.27
2.25
2.24
2.23
2.22
2.21
2.20
2.19
2.17
2.15
2.13
2.10
2.08
2.06

4
2.85
2.71
2.62
2.55
2.50
2.47
2.44
2.41
2.39
2.37
2.36
2.34
2.33
2.32
2.31
2.30
2.28
2.25
2.23
2.11
2.18
2.16
a = .05
5
2.98
2.83
2.73
2.66
2.60
2.56
2,53
2.50
2.48
2.46
2.44
2.43
2.42
2.41.
2.40
2.39
2.36
2.33
2.31
2.28
2.26
2.23
6
3.08
2.92
2.82
2.74
2.68
2.64
2.60
2,58
2.55
2.53
2.51
2.50
2.49
2.46
2.47
2.46
2.43
2.40
2.31
2.35
2.32
2.29

7
3.16
3.00
2.89
2.81
2.75
2.70
2.67
2.64
2.61
2.59
2.57
2.56
2.54
2.53
2.52
2.51
2. 48
2.45
2.42
2.39
2.37
2.34
8
3.24
3.07
2.95
2.87
2.81
2.76
2.72
2.69
2.66
2.64
2.62
2,61
2.59
2.58
2.57
2.56
2.53
2.50
2.47
2.44
2.41
2.38
9
3*. 30
3. 12
3.01
2.92
2.86
2.81
2.77
2.74
2.71
2.69
2.67 '
2.65
2.64
2.62
2.61
2.60
2.57
2.54
2.51
2.48
2.45
2.42
a- .01
1
3.37
3.14
3.00
2.90
2.62
2.76
2.72
2.68
2.65
2.62
2.60
2.58
2.57
2.55
2.54
2.53
2.49
2.46
2.42
2.39
2.36
2.33
3
3.90
3.61
3.42
3.29
3.19
3.11
3.06
3.01
2.97
2.94
2.91
2.88
2.86
2.84
2.83
2.81
2.77
2.72
2.68
2.64
2.60
2.5e
3
4.21
3.88
3.66
3.51
3.40
3.31
3.25
3.19
3.15
3.11
3.08
3.05
3.03
3.01
2,99
2.97
!.92
2.87
2.82
2.78
2.73
2.66
4
4.43
4.07
3.63
3.67
3.55
3.45
3.38
3.32
3.27
3.23
3.20
3.17
3.14
3.12
3.10
3.08
3.03
2.97
2.92
2.87
2.82
2.77
5
4.60
4.21
3.96
3.19
3.66
3.56
3.48
3.42
3.37
3.32
3,29
3.26
3.23
3.21
3.18
3.17
3.11
3.05
2.99
3.94
2.89
2.84
6
4.73
4.33
4.07
3.88
3.75
3,64
3.56
3.50
3.44
3.40
3.36
3.33
3.30
3.27
3.25
3.23
3.17
3.11
3.05
3.00
2.94
2.60
7
4.85
4.43
4.15
3.96
3.82
3.71
3.63
3.56
3.51
3.46
3.42
3.39
3.36
3.33
3.31
3.29
3.22
3.16
3.10
3.04
2.99
2.93
8
4.94
4.51
4.23
4.03
3.89
3.78
3.69
3.62
3.56
a. si
3.47
3.44
3.41
3.38
3.36
3.34
3.27
3.21
3.14
3.08
3.03
2.97
9
5.03
4.59
4.30
4.09
3.94
3.83
3.74
3.67
3.61
3.56
3.52
3.48
3.45
3.42
3.40
3.38
3.31
3.24
3.18
3.12
3.06
3.00
                    Vrom: Miller, 1981
                                      341

-------
 :.  COMPUTER  CALCULATIONS

      This  computer program incorporates  two  analyses:  an  analysis  of
      nce  (ANOVA),  and a multiple comparison  of  treatment  means with the
 :ontrol mean (Dunnett's Procedure).   The ANOVA  is  used to obtain the  error
 alue.  Dunnett's  Procedure indicates which  toxicant concentration means  (if
|iny)  are  statistically different from the control  mean at the 5% level of
 significance.   The program also provides the minimum difference between the
 Control and  treatment means that could be detected as  statistically
 lignificant, and tests the validity  of the homogeneity of variance
 Assumption by Bartlett's  Test.   The  multiple comparison is based on Dunnett,
    W.,  1955, "Multiple Comparison Procedure  for Comparing Several  Treatments
 nth  a  Control," J.  Amer.  Statist. Assoc.  50:1096-1121.

1.2  The  source code" for  the Dunnett's program  is  structured  into  a series
 )f  subroutines, controlled by a driver routine.  Each  subroutine has  a
[specific  function  in the  Dunnett's Procedure, such as  data input,
 ransforming the data, testing  for equality  of  variances,  computing P
 values, and  calculating the one-way  analysis of variance.

|2.3  The  program compares up to seven toxicant  concentrations against the
[controls  and can accommodate up to 50 replicates per concentration.

12.4  If the  number of replicates at  each toxicant  concentration and control
[are not equal,  Bonferroni's t-test is performed instead of Dunnett's
procedure  (see  Appendix D).

|2.5  The  program was written in IBM-PC FORTRAN  (XT and AT)  by D, L. Weiner,
^Computer  Sciences  Corporation,  26 W.  Martin  Luther King Drive, Cincinnati,
      45268.   A  compiled version of the program  can be  obtained from Computer
^Sciences  Corporation by sending a diskette with a  written request.

      Data  Input and  Output

(2.6.1   Data  on  the proportion of surviving mysids  (Mysidopsis bahia), from a
.survival,  growth and fecundity  test,  listed  in  Table C.6,  below, are  used to
[illustrate the  data  input and output  for this program.

[2.6.2   Data  Input

  6.2.1  When the  program is entered, the user  has the following options:

    1.  Create a data file
    2.  Edit  a data file                                      ,
    3.  Perform  ANOVA (analysis)  on existing  data set
    4.  Exit  the program
                                      342

-------
   TABLE C.6  SAMPLE DATA FOR DUNNETT'S PROGRAM.
              MYSIDS.
PROPORTION OF SURVIVING
                                            Replicate
 Treatment
| Control
I 50-
I TOO.
f 210.
450.
0
0
0
0
0.
0.
0.
1.
0.
80
80
60
00
00
0.
1.
1.
0.
0.
80
00
00
80
20
1.00
0.80
1.00
0.20
0.00
1.00
0.80
1.00
0.80
0.20
1.00
1.00
1.00
0.60
0.00
1.00
1.00
0.60
0.80
0.00
1.00
0.80
0.80
0.80
0.00
0.80
1.00
0.80
0.80
0.40
2.6.2.2  When Option 1 (Create a data file) is selected, the program prompts
the user for the following information:

    1. Number of groups, including control
    2. For each group:
       - Number of observations
       - Data for each observation

2.6.2.3  After the data have been entered, the user may save the file on a
disk, and the program returns to the main menu (see below).

2.6.2.4  Sample data input is shown below.
                                       343

-------
             MAIN MENU  AND DATA  INPUT
 l)  Create a data file
 2)  Edit  a data file
 3)  Perform aNWA on existing data set
 4)  Stop

Your choice ?  l

Nurrber of observations  for group  1 ? 8

Biter the data for group 1 one observation at a time.

NO.   1? 0.80

NO,   2? 0.80

NO.   3? 1.00

NO,   4? 1.00

NO.  5? 1.00

NO.  6? 1.00

NO.  7? 1.00

NO.  8? 0.80

Niirtoer of observations for group  2 ? 8



Do you wish to save the data on disk ?y

Disk file for output ? sanple
                            344

-------
 2.3.3  Program Output

 2.3.3.1  When Option 3  (Perform ANOVA on existing data set) is selected  from
jthe main menu, the  user is asked to select  the  transformation desired, and
indicate whether they expect the means of the test groups to be less or
greater than the mean for the control group (see below).
         1)  Create a data file
         2}  Edit  a data file
         3)  Perform AHOVA on existing data set
         4)  Stop

        Your choice ?  3
        File name ?  sample
         Available Transformations
            l)  no transform
            2)  square root
            3)  loglO
            4)  arcsine square root

        Your choice ? 4
          Dunnett's test as implemented in  this program is
          a one-sided test. You must specify the direction
          the  test is to be run; that is, do you expect the
          means  for the test groups to be less than or
          greater than the mean for the control group mean.

         Direction for Dunnetts test : L=less than, G=greater than ?  1
                                        345

-------
12.3.3.2  Summary  statistics  for the Yaw  and transformed data,  if.
  pplicable,  the ANOVA  table,  results of  Bartlett's  Test,  the results  of
Ithe  multiple comparison procedure  and the minimum detectable difference
 are  included in the program  output.
                           Stannary Statistics for Raw Data
               Group
                                 Mean
                                               s.d.
cv%
1




= control
2
3
4
5
8
8
8
8
8
.9250
.9000
.8500
.7250
.1000
. 1035
.1069
.1773
.2375
.1512
11.2
11.9
20.9
32.8
151.2
                Group
                             Suimary Statistics and ANCWA

                          TransfornBtion = Arcsine Square Root

                          n       Msan         s.d.           cv%
1




= control
2
3
4*
5*
8
8
8
8
8
1
1
1
1

.3969
.3390
. 2837
. 0570 .
.2015
.2400
.2478
.3181
.3066
.2863
17.
18.
24.
29.
142.
2
5
8
0
1
              *) the mean for tnis group is significantly less than
                 the control mean at alpha = 0.05 (l-sided) by IXtnnett's test
              Minununt detectable difference for Dunnett's test =        -.316663
              This corresponds to a difference of       -.192003 in original units
              Tnis difference corresponds to   -19.79 percent of control
              Between groins sum of squares «        7.826581 with  4 degrees of freedom.

              Error mean square =         .079230 with 35 degrees of freedom.

              Bartlett's test p-value for equality of variances =   .934
                                           346

-------
12.4  Listing  of Computer  Program for Dunnett's Procedure.
   Sstorage:2
   c
   c
   c
   c
   c
   c
   c
   c
   c
   c
   c
   o
   c
   c
   c
   c
   c
   c
   c
   c
   c
   c
   c
   c
   c
   c
   c
   c
   c
   c
   c
   c
   c
   c
   c
   c
   c
   c
   c
   c
   c
EPA Dunnett's test program. : Version 1.1  by D.L.Weiner, 9/10/87
Written for IBM PC and full compatibles.  May require modifications
for other systems.  This version compiled using Microsoft FORTRAN
compiled, V3.3.

 Driver program for D.exe

 This program does the following :
   1) creation and/or editing of ascii files in the following
      format (data are for example  purposes only):
                           The first column denotes the group and
                           the second column the data value.   Note
                           that it is assumed that group 1 is the
                           control.   Delimiters can be commas or
                           blanks.   Groups can have unequal sample
                           sizes.
               1,  23.4
               1,  17.6
               1,  59.0
               2,  44.1
               2,  50
               2,  51.7
               3,  49.8
               3,  39.0
               3,  56.2
               4,  73.4
               4,  64.9

  2) temporary transformation of the data for analysis purposes
     the transformed data are not permanently saved
  3) a one-way ANOVA
  4) Bartlett's test for equality of variances
  5) Dunnett's test to compare the comtrol mean vs. each of the
     test group means if the sample sizes are equal; otherwise
     simple t-tests (using a pooled error term) are done with
     Bonferroni adjustment of p-value.

PROGRAM RESTRICTIONS
    * number of groups must be between 2 and 8 (inclusive)
    * number of observations per group must be between l and 50


Version 1.1  : D.L.Weiner,  July,1987

  program d
  implicit real*8 (a-h,o-z)                     .
  real*8 mdd
  character*79 title
  character*! ans
  dimension wk(2000),  n(20),  y(400),  ip(20)
  dimension wkt(2000),  nt(20),  yt(400)
  data n/20*0/, nt/20*0/

  variables :  ng=number of  groups
              n(g)=array of  untransformed sample sizes
              ntot=sura of n(i),  i=l,g
              y(ntot)=array  of  response data
                                      347

-------
4  Listing of Computer Program for  Dunnett's Procedure (Continued).
                      iunit=unit number for output
                      nt(g)=array of sample sizes for transformed data
                      yt(ntotal)=array of transformed data
                      ntott=sum of nt(i), i=l,g
                      title=user specified title for output
                      ssb,ssw=between and within group sum of squares
                      sst=total corrected sum of squares
                      ems=error mean square with idf degrees of freedom
                  .    wk()=work array - contains means and variances
                      wkt()=work array - contains transformed means and va
                          work arrays are also used for other purposes  als
                      mdd=minimum detectable difference

       see source code for individual subroutines for additional documenta


          call input(ng,n,ntot,y,iunit,title)
          call trans(ng,n,ntot,y,nt,yt,ntott,inum)

      summarize raw data if transformation requested in addition
      to summarizing raw data

          idf = ntott - ng

          if(ng.lt.2.or,idf-It.5)   then
            write(*,'(/a/)'}  '  Not enough data values for  analysis,'
            goto 10
          endif
          if(ng.gt.8)   then
            write(*,'(/a/)')  '  Too many groups for analysis.'
            goto 10
          endif

          if(inum.ne.1)  call  oneway(ng,n,ntot,y,sst,ssb,ssw,wk)
          cal1 oneway(ng,nt,ntott,yt,sst,ssb,ssw,wkt)
          ems = ssw / idf
          call eqvar(ng,nt,wkt(ng+l),p)

    p is p value for Bartlett's test  :  if p >  1 then the test  couldn't
    be run as one or more of  the variances are zero

          call dunnet (ng, idf,ems,wkt (1) ,nt,wkt (2*ng+l) , iside, ip,utdd)

     call  summary to summarize  raw data  if transformation  requested
     in addition to summarizing raw data :  next to last arg =  0
     means no ANOVA summary - n, mean,  sd only

     summary is called  twice  -  once for  screen output  and  once for
     printer or disk output

          call els

     summarize raw data here  (if analysis is on transformed data)
     note that ssb,ssw,p,ip,iside  are not used in this call (dummy)
                                       348

-------
,4   Listing of  Computer Program for Dunnett's Procedure (Continued).
            i f ( inura . ne . 1 )  then
            call summary (ng,n,ntot,wk(l) ,wk(ng-H) , ssb,ssw,p, ip, iside,
         &      title, 0,0, mdd)
            pause '   *
            call summary (ng,n,ntot,wk(l) ,wk(ng+l) , ssb,ssw,p, ip, iside,
         &      title, iunit.o, add)
            endif
    c
    c  summarize transformed data here
    c
            call summary (ng,nt,ntott,wkt (1) ,wkt (ng+1) , ssb, ssw,p, ip, iside,
         &      title, 0, inum, mdd)
            call summary (ng,nt,ntott,wkt(l) ,wkt(ng+l) , ssb, ssw.p, ip, iside,
         &      ti tie, iunit, inum, mdd)
10
             write(iunit, ' (lx,al) ' )  char(12)
             close(iunit)
             write (*,' (/a\) ')  '  Do you wish to restart  the program ?  '
             read(*, ' (al) ')  ans
             if (ans.eq. 'Y ' .or. ans. eg. 'y ' )  goto 1
             if (ans.ne. 'Nf .and. ans. ne. 'n')  goto 10

             stop 'Normal  ending.1
             end
                                       349

-------
;2.4  Listing  of Computer  Program for  Dunnett's Procedure.
                                 data input routine

                             variable     type      description
g
ntot
n(g)
y(ntot)
iunit
title
i-2
12
12
r8
12
a79
number og groups
total | of obs.
| obs. per group
data values
unit for output
title
$storage:2
c
c       input.for
c
c        on output
c
c
c
c
c
c
c
c
        subroutine input(g,n,ntot,y,iunit,title)
        implicit real*8  (a-h,o-z)
        dimension y(l),n(l)
        integer*2 g
        character*64 fname
        character*79 title
        character*! ans
        logical  iochk
c
        call els
        call gotorc(8,l)
        write(*.'(a/,a/,a/,a/,a)')
      &     *******

      &  \     *                 EPA
      &  ,     *
      &  ,     *
      &  t     ******
        write(*,'(/a\)')  ' Title ?
        read(*,•(a79)')  title
 2       write(*,'(/a\)')  ' Output to  printer or disk file ?
        read(*,•(al)')  ans
       if  (ans.ne.'p1.and.ans.ne.'P1.and.ans.ne
                                          Dunnett''s
                                        Version 1.1
Program
 *
*'
* i
                                                     d'.and.ans.ne.'D') then
                 write(*,*)  ' Please respond with a p or a d  '
                 pause
                 call els
                 goto 2
              end if
              if(ans.eq.'p'.or.ans.eg.'P') then
                 iunit =20
                 open(iunit,file='prn')
                 goto 5
              endif
                 iunit = 10
              write(*,'(/a\)')  '  Disk file for output ?  '
              read(*,'(a64)') fname
              inquire(file=fname,exist-iochk)
              if(iochk) then                                    ,    _  .
               write  (*,'(/a\)')  '  File  already  exists, overwrite  it ?
               read(*,'(al)') ans
               if(ans.eq.'N'.or.ans.eg.'n') goto 3
                   if(ans.ne.'Y1.and.ans.ne.'y')  then
                                         350

-------
£.4   Listing of  Computer Program for Dunnett's Procedure  (Continued).
c
5




6

call cl
write (*
write ( *
write (*
write (*
write (*

3
*)
*>
*)
*)
'(
    C
    c
    c
    15

    16
   22
   20
                 write(*,*) '  Please answer yes or no. '
                 goto 4
                 endif
            endif
            open(iunit,file=fname,access='sequential',status='new'}
                 1} Create a data file1
            *} ' 2) Edit a data file'
                 3) Perform ANOVA on existing data set1
                 4) Stop'
            1  (/a\)')  '  Your choice ? '
    read(*,'(al)') ans
    i f(ans.ne.'1'.and.ans.ne.' 2'.and.ans.ne.'3'.
 &      and.ans.ne.'4')  then
      write(*,'(/a/)')  '  Please enter a number 1,2 or 3*
      goto 6
    endif
    if(ans.eq.'4') stop 'Normal Ending.1
    if(ans.eg.'3'} goto 30
    if(ans.eg.'2') goto 100

input from keyboard here

     call inkb(g,n,ntot,y)
     call els
     write(*,*(/a\)')  •  Do you wish to  save the  data on disk ?'
     read(*,'(al)') ans
     if (ans.eq. fKr .or. ans. eq.'^n1)  then
        write(*( »(/a,a\) ') •  Are you sure ?  Data will Jse lest
 &                         *  (Y or N) ? '
        read(*,'(al)')  ans
        if(ans.eq.'Y'.or.ans.eq.'y') goto 5
        goto 16
     endif
     if(ans.ne.'y'.and.ans.ne.'Y')  then
       write(*,•(/a)')  •  Please respond yes or no.1
       goto 16
     endif
    write(*,'(/a\)')  '  Disk file  for output ?  '
    read(*,'(a64)') fname  •
    inquire(file=fname,exist=iochk)
    if(iochk)  then
      write (*,'(/a\)')  '  File  already  exists, overwrite it ?  '
      read(*,'(al)')  ans
      if(ans.eq.'N1.or.ans.eg.'n')  goto 16
    endif
    iunitl =  11
    open(iunitl,file=fname,access='sequential1,status='new')
    1 = 0
    do 20 j=l,g
       do 22 k=l,n(j)
       1  = 1+1
       write(iunitl,'(lx,i4,2x,a,2x,f!8.6)') j,  ',*
     continue
                                         351

-------
2.4  Listing of Computer Program for  Dunnett's Procedure (Continued)
    25
    c
    c
    c
    30
    100
   10

   20


   22
   25

   28
     close(iunitl)
     goto 5

input from disk file here

    continue
    call readf(g,n,ntot,y,iflag,fname}
    if(iflag.eq.O) return
    goto 5
    continue
    call readf(g,n,ntot,y,iflag,fname)
    if(iflag.eq.O) call  edit(g,n,ntot,y,fname)
    goto 5
    end
            inkb.for

            on output
                    .  keyboard  input

                   variable     type
description
g
ntot
n(g)
y(ntot)
i2
i2
i2
re
number og groups
total f of obs.
# obs. per group
data values
    subroutine  inkb(g,n,ntot,y)
    implicit  real*8(a-h,o-z)
    dimension y(l),n(l)
    integer g
    call els
    write(*,*') ' Number of observations for group ',
s       i,'  ?  '
    read(*,*,err=25) n(i)
    if(n(i).le.0.or.n(i),gt.50)  then
     write(*,*)  ' The number of observations per group must be '
&    '1 to 50*                                       *          '
     goto 22
   endif
   goto 28
   write(*,»(/a/)') ' Invalid number.  Please reenter.1
   goto 22
   write(*/'(/a,i2)a) ')  '  Enter the data for group M,
&      '  one observation  at a time.'
   do 100  j=l,n(i)
                                         352

-------
,4   Listing of  Computer Program for Dunnett's Procedure  (Continued)
 30
 35

 100
 200
 30
write(*,'(/a,i2,a,\)') '  NO. ',;),'? '
read(*,*,err=35) ynum
1=1+1
y(l) « ynum
goto 100
write(*('(/a/)') '  Invalid number. Please reenter.•
goto 30
continue
continue
ntot m i
return
end
         readf.for
          on output :
                   read a data file
                variable
type
description
g
ntot
n(9)
y(ntot)
iflag
fname
i2
i2
12
r8
12
a64
number og groups
total # of obs.
# obs. per group
data values
0 = ok read, 1 = not
file that was read in




ok

c
c
c
c
c
c
c
c
c
c
c
c
        subroutine readf(g,n,ntot,y,iflag,fname)
        implicit real*8 (a-h,o-z)
        integer g
        character*64 fname
        character*! ans                    :  ' '••
        logical iochk                      x
        dimension y(l),n(l)
        call els
        iflag = o
        write(*,'(/a\)') '  File name  ? '
        read(*,'(a64)')  fname
        if(fname.eg.'  ') goto 31
        inguire(file=fname, exist=iochk)
        if(iochk) goto 50
31    write(*,*)  *  The file you specified does not exist,  or
                                                              you  need1
       write(*,*J  »  to specify a different disk as part of  the  name
         write(*,'(/a\)')  '  Do you wish to reenter the name ?  •
         read(*,'(al)')  ans
         if (ans,eg.fY'.or.ans.eg.'y')  goto 30
         iflag =  1
         return
 50      continue
 c    begin file  read
         iunit -  30
         open(iunit,file-fname,access='seguential',status='old'
         readtiunit^^end^lOO)  igrpr  x
 c i is obs#,  ig  is  group*,  nn counts #obs per group

         ig~- 1
         nn = 1
         ilag = ig
                                       353

-------
I

2.4  Listing  of Computer Program for Dunnett's Procedure  (Continued).
     60
 iglag = igrp
 read(iunit,*,end=100,err=150)  igrp, x
 if(iglag.eg.igrp)  nn  =  nn +  1
 if(iglag.ne.igrp)  then
      n(ig)  - nn
      ig = ig +  1
                  iglag  =  igrp
                  nn = 1
             end if
    100
    150
y(i) = x
if (nn.gt.50) then
  write (*, ' (\a,i2,a\) ')  ' Too many observations for croup
   1  Max =  50 '
  iflag = l
  pause
  return
endif
goto 60
continue
n(ig) = nn
ntot = i
g «• ig
close(iunit)
return
continue
write (*, l (a,i2,a,a) ') '  There is an error on line (,
            pause
            iflag = l
            return
            end

            edit.for

             on input
                'your data file. Please correct it.'
                                                                    of
                  file edit routine
               variable
type
                                                  description
g
ntot
n(g)
y(ntot)
fname
12
12
12
r8
364
number og groups
total # of obs.
# obs. per group
data values
file to be edited
            subroutine edit(g,n,ntot,y,fname)
            implicit  real*8(a-h,o-z)
            dimension y(l),n(l)
            integer g
            character*64  fname
            call  els                   .  ;:'
            fuzz  = l.e-20
            iunit «= 30
            open(iunit,file=fname,access='sequential'.status='old()
            wnte(*,'(/a\)') ' Edit values for which group ? '
            read(*,*(err=10) ig
                                        354

-------
2.4  Listing of Computer Program for  Uunnett's Procedure (Continued).
    10
    100
    102
    c
    105
    110
    120
   130
   140
   150
                               ') * The following values are for group ',ig
    if(ig.ge.l.and.ig.le.g) goto 100
    writef*,'(//a,i2,/)')  ' Please respond 1 - '  g
    pause
    goto 1
    continue
    call els
    call gotorc(0,o)
    writef*, ' (a,i:
    call gotorc(5,6)
    loff m o
    if (ig.eq.l) goto 105
compute offset
    do 102 i=l,ig-l
    loff = loff 4 nfi)

    do 110 j=l,nfig)
    irow =5+ (j-i) /  4
    k = mod(j,4)
    iffk.eq.l)  icol = 0
    if(k.eq.2)  icol = 20
    if(k.eq.3)  icol = 40
    iffk.eq.oj  icol = 60
    call  gotorc(irow,icol)
    writef*,'(fis.6,\)')  y(loff  4  j)
    call  gotorc(23,0)
    write(*,'
-------
|2.4  Listing of  Computer Program for Dunnett's  Procedure (Continued)
     200     continue
             write(*,'(/a\)')  '  Do you wish to save the changes ? '
             read(*,'(al)')  ans
             if(ans.eg.'n1.or.ans.eg.'N') then
               close(iunit)
               return
             endif
             if(ans.ne.'y1.and.ans.ne.*Y') goto 200
             close(iunit)
             open(iunit,file=fname,access**'seguential',status='new')
             1-0
             do 250 j«l,g
                do 252 k«l,n(j)
                1 = 1 +  1
     252        write(iunit,'(Ix,i4,2xla(2x,f18.6)') j, ',' ,y{l)
     250      continue
              close(iunitl)
              return
              end
                                         356

-------
2.4  Listing of Computer  Program for Dunnett's Procedure  (Continued)
     $storage:2
     c
     c   cursor positioning, goto  (irow,icol}
     c
           subroutine gotorc  (irow,icol}
           character*! dummy

           if  (irow.lt.0.or.irow.gt.24) irow = 0
           if  (icol.It.0.or.icol.gt.79) icol «* 0
           read  (dummy,'(Ix)')
           call  locate (0,0,ier)
           write (*,'(\)'J
           call  locate (irow,icol,ier)
           read  (dummy,'(Ix)')
           return
           end
 0 ... 79

24
                                        357

-------
2.4  Listing of Computer  Program for  Dunnett's Procedure (Continued).
Sstorage:2
c
c trans. for - computes transformed data
c
c on input : variable type description
C g
c n()
c ntot
1 ° y()
1 c
I c on output : variable
! c 	 	 	
c nt()
c yt()
c ntott
c inum
12
12
i2
r8
type
12
r8
r8
12
number of groups
n values for each group
total # of obs.
data values
description k
n values after transformati It
data values after transform Jj|
ntot after transformations jm
transformation number i$|
            subroutine trans(g, n,ntot,y,nt,yt,ntott,inum)
            implicit real*8(a-h,o-2)
            integer g
            dimension n(l), y(l), nt(l),  yt(l)
    c
            iflag = 0
    5       call els
            write(* *)   Available Transformations'
            write(* *)       l)   no transform1   -  ,
            write(* *)       2)   square root'
            write(* *)       3)   logio1
            write(* *)       4)   arcsine square root'
            write(* l(/a\)'} «  Your choice ? •
            read(*,*,err=50) inum
            if(inum.It.l.or.inum.gt.4)  goto  50
            goto 60
    50      write(*,*) •  Please  answer  1-4 •
            pause ' *
            goto 5
    60      k = 0                            T "V
            kk « 0                           : f-i  "-?•
            ntott «= 0                           :
            do 52 1=1, g                        )'
               nt(i) = o
               do 54 j=l,n(i)
                   k = k  + l
                   goto(70,72,74,76),inum
    70      temp = y(k)
            goto 53
    72      if (y(k).lt.O.dO) then
               iflag - 1
               goto 54
            endif
            temp m dsgrt(y(k))
            goto 53
    74      if (y(k).le.o.dO) then
              iflag  «  l
                                        358

-------
 Listing of Computer Program for Dunnett's  Procedure  (Continued).
          goto 54
        endif
        temp = dloglO(y(k))
        goto 53
76      if (y(k).lt.O.dO.or.y(k).gt.l.dO)  then
          iflag - 1
          goto 54
        endif
        temp = dasin(dsqrt(y(k}))
53      kk - We + 1
        nt(i) - nt(i) + 1
        ntott - ntott + 1
        yt(kk) - temp
54      continue
52      continue
c
        if(iflag.gt.O) then
          write(*,'(/a)')
     &   ' One or more data values could not be transformed.  These
        write(*,'(a/)r)  ' values will not be included in the  analyses.
          pause  '  '
          endif
c
c   check to see if each group has at least 1 observation
c
          do 100 i«l, g
100       if(nt(i).le.O) goto 110
          return
110       call els
       write(*,'(a,a,i2J') • After transformation,  all of the values',
     &    « are missing  for group ', i
          write(*,'(/a)'} ' ANOVA cannot be performed. Program  ending,
          stop  ' l
          end
                                      359

-------
2.4  Listing of Computer Program for  Dunnett's Procedure (Continued).
$storage:2
c
c oneway. for
c
c on input :
c
c
c
c
c
c
c on output :
I c
c
c
c
c
c
c
computes
variable
g
"0
ntot
y()
wk()
variable
wk(l)~wk(g)
wk(g+l)-wk(g+g)
sst
ssb
ssw
oneway
type
12
12
12
r8
r8
type
r8
r8
r8
r8
rs
anova
description
number of groups
n values for each group
total sample size
data values (possible tr
work array (dimensioned
. description
group means
group variances
corrected total sum o
between group sum of
within sum of squares

ans
2*g
f s
sgu
    10
    c
    30
    C
    c
    c
    c
    c
    c
    20
    c
     subroutine oneway(g,n,ntot,y,sst,ssb,ssw,wk)
     implicit real*8  (a-h,o-z)
     integer*2 g
     dimension n(l), y(l), wk£l)

     sy = 0.0
     syy = 0.0
     sst ~ 0.0
     ssb = 0.0
     ssw =0.0
     k = 0
     temp =0.0             '•••:

     do 10 i=l, 2*g
     wk(i)=0.0

     do 20 1=1, g
             do 30 j=l,n(i)
             k = k + 1
             sy = sy + y{k)
             syy = syy + y(k)  * y(k)
             wk(i) » wk(i) + y(k)
             wk(g-t-i)  = wk(g+i)  + y(k)*y(k)

compute and store 1th group variance  in wk(g+i)

     wk(g+l)  - wk(g+i) - wk(i)  * wk£i)  / n(i)
     if(n(i).gt.l) wk£g+i) * wk(g4i)  /
     temp = temp + wk(i) * wk(i)  /  n£i)

compute and store ith mean in  wk(i)

     wk(i)  - wk(i) / n(i)

     sst = syy - sy * sy / ntot
            ssb = temp  -  sy * sy / ntot
            ssw •= sst - ssb

            return
            end
                                         360

-------
,4   Listing of  Computer Program for Dunnett's  Procedure  (Continued).
  Sstorage:2
  c
  c       eqvar.for
  10
                     computes tests  for equality of variances
           on input :
variable
g
var(g)
variable
P
type
12
12
12
r8
type
r8
description
number of groups
n values for each group
array of variances
data values (possible trans
description
Bartlett's test p-value
           on output
 subroutine eqvar{g,n,var,p)
 implicit real*8(a-h,o-z)
 integer g
 dimension var(l),  n(l)

 vmin = var(l)
 vmax = var(l)
 imin « 1
 imax «= 1
 vsum - var(l)
 sse  « (n(l)-l)  * var(l)
 idf  » n(l)  -  l
 cl - O.do
 c2 = O.dO
 if(var(l).le.l.d-10) goto 100
 cl = (n(l)-l) * dloglO(vard))
 if ((n(i)-l).gt.O) c2 = l.dO /  (n(l)-l)
 do 10 i=2,g
 vsum « vsum + var(i)
 sse  = sse + (n(i)-l) * var(i)
 idf  = idf + (n(i)  - 1)                  : . :.
 if(var(i).le.l.d-10) goto 100
 cl - cl  +  (n(i)-l) * dlogiofvar(i))
 ifUn(i)-l).gt.O)   c2 = c2 + l.dO / (n(i)-l)
 if (var(i).le.vmin) then
     vmin =  var(i)
     imin =  i
 endif
 if (var(i).gt.vmax) then
    vmax = var(i)
     imax =  i
 endif
continue
 f = vmax / vmin
 idfl - n(imax) - l
 idf2  = n(imin) - 1
call  pvalue (3,2,f,idfl,idf2,pi)
    note pi = p-value for F max test
                                       361

-------
2.4  Listing  of Computer Program for Dunnett's Procedure (Continued)
             C  -  l.dO +  (l.dO /  (3*(g-l))J * (c2 - l.dO/idf)
             chi  »  2.303 *  (idf  * dloglO(sse/idf)  - cl)
             chi  =  chi / c
             idfl = g-l
             idf2 - 0
             call pvalue (4,2,chi,idfl,idf2,p)
             return
     100      continue
     c    set  p  to 2 (a flag) if l or more variances =  0
             p  =  2.dO
             return
             end
                                       362

-------
•
[2.4   Listing of  Computer Program for Dunnett's Procedure  (Continued).
variable
itype
value
idfl
idf2
variable
P
type
i2
12
r8
12
i2
type
r8
description
1-t, 2=z, 3=f , 4=chi square
1=1 sided, 2=2 sided
value of test statistic
1st degrees of freedom t,z,
2nd degrees of freedom f
description
p-value
    Sstorage:2
    c
    c       subroutine pvalue.for - compute p values
    c
    c
    c        on input :
   -C
    c
    c
    c
    c
    c
    c
    o        on output :
    c
    c
    c
            subroutine pvalue(itype,i!2,value,idfl,idf2,p)
            implicit real*8 (a-h,b-z)
            goto (10,11,12,13) itype
   '10      t = value
            idf « idfl
            aa=idf/2.do
            bb=0.5dO
            xx-l.dO/(l.d(H(t**2)/idf)
            bta=beta(xx,aa,bb)
            if(i!2.eq.l.and.t.gt.0.dO)  p-bta/2.dO
            if(i!2.eq.l.and.t.lt.0.dO)  p=(l.do-bta/2.dO)
            if(i!2.eq.2)  p = bta
            return
    11      z = value
            aa=l.dlO/2.dO
            bb*=0.5do
            xx=l.dlO/(l.dl(H(z**2))
            bta=beta(xx,aa,bb)
            if(il2.eq.l.and.z.gt.0.dO)  p=bta/2.dO
            if(i!2.eq.l.and.z.lt.0.dO)  p=(l.dO~bta/2.dO)
            if(i!2.eq.2)  z = bta
            return
    12      f - value
            aa=idf2/2.dO
            bb=idfl/2.dO
            xx=l.dO/(l-dO+(idfl*f)/idf2)
            p=beta(xx,aa,bb)
            return
    13      chi  »  value
            idf  =  idfl
            aa=l.dlO/2.dO
            bb=idf/2.dO
            xx=l.dlO/(1.dlO+chi)           ;
            p-beta(xx,aa,bb)
            return
            stop
            end
                                         363

-------
Ri
K.4  Listing of Computer Program  for Dunnett's  Procedure  (Continued)
10
11
12

13
14
            double precision function beta(xx,aa,bb)
            implicit real*8(a-i,k-z)
            integer*2 j
            ier » 0                        .
            BETA=1.0
            IF (XX - 1.0) 1,30,30
            BETA=0.0
            IF (XX) 30,30,4
            A=AA
            B=BB
            x=xx
            LO=DLOG(X)
            Ll-DLOG(l.O-X)
            M=DLGAMA(A) + DLGAKA(B) - DLGAMA(A+B)
            IF (A-1.5) 6,7,7
            BETA=BETA+DEXP(A*LO+B*L1-M)/A
            M=M+DLOG(A)-DLOG(A+B)
            GO TO 5
            IF (8-1.5) 8,9,9
            BETA=BETA-DEXP(A*LO+B*L1-M)/B
            M=M+DLOG(B)-DLOG(A+B)
            B-B+1.0
            GO TO 7
            Y-1.28155156553942DO
            L=(Y*Y-3.0)/6.0
            H=(2.0*A-1.0)*(2.0*B-1.0)/(A+B-1.0)
            Z=(L+(5.0*H-4.0)/(6.0*H))*(A-B)/(A+B-1.0}*2.0/H
            L=DSQRT(K+L)/H
            CUT=-V*L-Z
            CUT=A/ (A+B*DEXP(2. Q*CUT) )
            IF (CUT-0.5) 12,12,10
            CUT=Y*L-Z
            CUT=A/ (A+B*DEXP(2.0*CUT) }
            IF (CUT-0.5) 11,12,12
            CUT=0.5
            H-1.0
            IF (X-CUT) 14,14,13
            H=LO                                   .
            LO=L1
            Ll-H
            x-i.o-x-
            H=A
            A=B
            B=H
            BETA-BETA+1.0
            H— 1.0
            EPS=1.0D-16
            M=DEXP (A*LQ+ (B-l . 0) *L1-M)/A
            X=X/(1.0-X)
            I>=0.0
            Y=l,0
            2=1.0
            DO 18 J=l,100
            I=J
                                          364

-------
,4   Listing of Computer  Program for Dunnett's Procedure (Continued).
   15
   20
   16
   18
   19
   30
   L0=(I-B)*(A+I-1.0)/(A+2.0*1-2.0)*X/(A+2.0*1-1.0)
   L1«(A+B+I-1. 0)*!/(A+2.0*1)*X/(A+2.0*1-1.0}
   L=L*LO+M
   Z=Z*LO+Y
   M=M*L1+L
   Y=Y*L1+Z
   IP (Z)  15,18,15
   S=L/Z
   10
   IF (T)  16,20,16
   IP (S)  18,19,18
   IF (DABS(S/T-1.0)-EPS)  19,19,18
   CONTINUE
   BETA-BETA+H*T
   RETURN
   RETURN
   END
   DOUBLE PRECISION FUNCTION DLGAMA(XX)
   DOUBLE PRECISION XX,ZZ,TERM,RZ2,DLNG
   IER=0
   zz=xx
   IF (XX-1.D10) 2,2,1
   IF (XX-1.D35) 8,9,9
   IF (XX-l.D-9) 3,3,4
   IER—1
   DLNG=-1.D38
   GO TO 10
   TERM=1.DO
   IF (ZZ-18.DO) 6,6,7
   TERM=TERM*ZZ
   ZZ=ZZ+1.DO
   GO TO 5
   RZ2=1.DO/ZZ**2
 DLNG =(ZZ-0.5DO)*DLOG(ZZ)-ZZ +0.9189385332046727 -DLOG{TERK}+
1(1.DO/ZZ)*(.8333333333333333D-1-(RZ2*(.2777777777777777D-2+(RZ2*
2(.7936507936507936D-3-(RZ2*(.5952380952380952D-3)))))))
   GO TO 10
   DLNG-ZZ*(DLOG(ZZ)-1.DO)
   GO TO 10
   IER=+1
   DLNG=1.D38
   DLGAHA»DLNG
   RETURN
   END
                                         365

-------
;2.4  Listing  of Computer Program for Dunnett's Procedure  (Continued).
    $storage:2
    c
    c       subroutine dunnet.for - compute p values
    c
    10

    20

    C
    c
    c
    22
             on input  i
                     variable
type
                          description
             on output  :
ng
idf
ems
mean(ng)
n(ng)
t<8,50)
iside
variable
ip(ng)
12
12
rs
ra
12
r8
12
type
12
number of groups 2<=ng<=8
degrees of freedom for erro
error mean square
array of means
n per each group
work array
0*trts lower, l=trts higher
description
0=NS, l=sig @ alpha=0.05
                               HDD
                                   r8
                           ip(l) = 0 => Dunnetts te
                           ip(l) = 1 => Bonferroni
                           min. detectable diff. in
                           original units
      note : the calling program must check to see that 2<=ng<=8 and
             that idf is >« 5

     subroutine dunnet(ng,idf,ems,mean,n,t,iside,ip,mdd)
     implicit real*8 (a-h,o-z)
     real*8 mean(l), mdd
     character*! ans
     logical iochk
     dimension t(7,49),  ip(l),  n(l)

  read in Dunnett's t values

     inquire(file=(dunnet.fil',exist=iochk)
     if(iochk) goto 10
   write(*,*}  '  The file containing  Dunnett t values is not on the'
   write(*,*)  '  default drive.   The  file name is DUNNET.FIL       '
   write(*,*)  *  Please copy it  over  to the default drive and rerun1
   write(*,*)  '  this program.                                     '
   stop '   '
     open (97,file-'dunnet.fil',status='old')
     do 20 j-1,49
     read(97,'(7f5.2)')  (t(i(j),i=l,7)
     close(97)

read in direction for the test
     call els
     write(* *)
            write(*
            write(*
            write(*
            write(*
             *)
             *)
             *)
             *)
Dunnett''s test as implemented in this program is
a one-sided test. Vou must specify the direction
the test is to be run; that is,  do you expect the
means for the test groups to be  less than or
greater than the mean for the control group mean.
                                         366

-------
30

c

c
c
c
         write(*,'(/a\)')

      &  read^aVranr16"8 tCSt : ^leSS than' Greater than 7  .

         "&£:£^S£M-tI't-"*-™-<»-'*'
                      ' Please respond L or G. '
          endif

          ifjans.eq.'I'.or.ans.eq.'L') iside=o
          if(ans.eq.'g'.or.ans.eq.'G') iside=l

          cmean « mean(l)


       check to see if sample sizes are equal :

          do 30 i=2,ng
            if(n(i)-n(i-i)) 100, 30, 100
          continue
          xn = i.do * n(l)


          denom = dsqrt(2,dO  * ems / xn)

       recover Dunnett's t  value

          icol  «  ng -  i

                              tficol
                                               if not,  do Bonferroni
         iffi
                       dunt = t(icol ,  49)
        ip(l)  = 0
        do  50  i=2,ng
        ip(i}=0
        if  (iside.eq.O) diff
        if  (iside.eq.l) diff
        stat =  diff / denom
        if  (stat.gt.dunt) ip(i)=i
                                cmean - mean(i)
                                meanfi) - cmean
 50
 c
 c  compute  MDD
 c                                ••...'•';
        mdd =  dunt  *  denom
o   *<  if(iside.eq.O) mdd = -i.do  * mdd
c   fixup if transformed is done by summary
        return                           J
100     continue
c   Bonferroni adjustment here
       alpha - 0.05do /  (ng -  i
       pctile « i.do  - alpha
       tval  - tinv (pctile,  idf)
       nl «  n(l)
       do iso i-2,ng
      ni
                                 367

-------
2.4  Listing  of Computer Program for Dunnett's  Procedure {Continued}
             denom - dsqrt(ems *  (l.do/nl + l.do/nij )
             if (iside.eq.O)  diff =  cmean - mean(i)
             if (iside.eq.l)  diff =  mean(i) - cmean
             stat « diff / denom
     150     if (stat.gt.tval)  ip(i)=l
     c
     c  compute HDD
     c
             denom = dsqrt(2.dO * ems / nl)
             mdd = tval * denom
             if(iside.eq.O) mdd = -l.dO * mdd
     c   fixup if transformed is  done by summary
             return
             end
                                        368

-------
'.4   Listing of  Computer Program for Dunnett's  Procedure  (Continued).
I $storage:2
1 °
I c summary . for
§ c
1 c on input :
Bj C
I C
1 C
I °
I- c
1 c
I c

I' <=
i c
1 <=
I c
1 c
1 c
1 c

I c
1 c
I c
[ c
I c
c
c
computes
variable
g
n()
ntot
mean ( )
var()
ssb
ssw
P
lp()



iside

title
iunit
inum


MDD


oneway
type
12
12
12
r8
r8
r8
r8
r8
12



12

a?9
12
12


rs


anova
description
number of groups
n values for each group
total sample size
data values (possible trans
work array (dimensioned 2*g
between group sum of squ
within sum of squares
Bartlett's test p-value
flag for Dunnett's test res
0=ns, l=sig
ip(l) - 0 => Dunnetts test
ip(l) = 1 => Bonferroni t-t
direction of Dunnett's test
0=lower, l=upper
title
unit # for output
0 means summarize raw data,
l=no trans, 2=sqrt,
3=loglO, 4=arcsine
min. detectable diff. for
Dunnett ' s test

            subroutine  summary(g,n,ntot,mean,var,ssb,ssw,p,ip,iside,title,
         &      iunit,inum,mdd)
            implicit  real*8  (a~h,o-z)
            real*8 mean(l), mdd, mddl
            Integer*2 g
            character*79 title
            character*20 tran(4)
            character*12 direct
            character*! pstar
            dimension n(l), var(l), lp(l)

            if(inum.eq.0} goto 2
            tran(l)='      None         '
            tran(2)='   Square Root     '
            tran(3)='     LoglO         '
            tran(4)-'Arcsine Square Root'
      convert  mdd to original units
            if  (inum.eq.1) then
                     tempc = mean(l)
                     tempt = mean (1) -f mdd
                     mddl = tempc - tempt
            end if
            if  (inun>,eq.2) then
                     tempc = mean(l) ** 2
                     tempt = (mean(l) •+• mdd) ** 2
                     mddl = tempc - tempt
            endif
                                        369

-------
2.4  Listing of Computer Program  for Dunnett's Procedure (Continued)
   30
            if (inura.eq.3)  then
                     tempo  = 10.0 **  (mean(l))
                     tempt  - 10.0 **  (mean(l)  +  radd)
                      mddl  * tempo -  tempt
            endif
            if (inum.eq.4)  then
                     tempo  = (dsin(aean(l)})**2
                     tern =  (mean(l) + mdd)
                     tempt  = (dsin(tem) )**2
                      mddl  = tempo -  tempt
            endif
            mddl  =  (dabs(mdd)/mdd)  *  dabs(mddl)
            pctrl =  100. do  * mddl  / dabs (tempo)

            check to Bee if all sample  sizes are equal
           iegn =  0
           do 3 i  = 2, g
           continue
           goto 5
           ieqn = 1
                             4,3,4
 if (iunit.gt.O)   write(iunit, ' (lx,al) ') char(12)
 write(iunit, ' (/,lx,a/) ')  title
 if (inum.eq.O) write(iunit, • (/,15x,a//) ')
   1  Summary Statistics  for Raw  Data1
 if (inum.ge. 1) then
   writefiunit, l (/(17x,a) f)
   1  Summary statistics  and AKOVA*
   writefiunit, • (/,13xfa, a/} ')  * Transformation =  '(tran{inum)
  endif
 write(iunit,*)
 1  Group      n         Mean           s.d.            cv% '
 writefiunit,*)

 if (var(l) .gt.O.dO) then        ~
     sd =  dsqrt(var(l))
     else
     sd =  o.dO
 endif
 if (mean(l) .ne.O.dO) then
     cv =  dabs(100.dO *  sd / mean(l))
     else  cv = O.dO
 endif
write (iunit,30) n(l) ,mean(l) ,sd,cv
 format ('  1 * control • ,lx, i2, 3x,fl2.4, lx,f 12. 4, 9x, f6. 1)

do 10 i»2,g
if (var(i).gt.O.dO)  then
    sd - dsqrt(var(i))
    else
    sd = O.dO
endif
if (mean(i) .ne.O.dO)  then
                                       370

-------
2.4  Listing of Computer Program  for  Dunnett's Procedure (Continued).
                cv - dabs(100.dO * sd / mean(i))
                else
                cv - o.do
            endif
            pstar='  '
            if(ip(i).eq.l)  pstar-'*'
            if(inum.ne.0}
             write (iunit,'(4x,i2,al,6x,i2,3x,fi2.4,lx,fl2.4,9x,f6.i)')
               i,pstar,n(i),raean(i),sd,cv
    c  don't  write out pstar if inum « o    -  summarize  raw data
    c
           if(inum.eq.0)

         &    write (iunit,'(4x,i2,7x,i2,3x,fi2.4,lx,fl2.4,9x,f6.1)')
         &      i,n(i),mean(i),sd,cv
    10      continue
           write(iunit,*)
         jt  »
           if(inum.eq.0) return


           if  (iside.eq.O) then
             direct='less than1
             else
             direct='greater than'
           endif

           if(ipfl).lt.l) then
           write(iunit,•(/a,a,/,a,a/)<)
        & '  *) the mean for this group is significantly 'direct,
        &      the control mean at alpha = 0.05 (l-sided)  by Duwiett'-s'
        & *  test'                                                        '
           else
                                 '
             )  the mean for this group is significantly 'direct,
               the control mean at  alpha « 0.05 (1-sided)  by a  t -  test1
               with Bonferroni adjustment of alpha level*                '
           endif


           if (iunit. eq.O)  pause '  •
           idf 1 = g - i
           idf2 = ntot - g
           f =  (ssb/idfi)  / (ssw/idf2)
           call pvalue(3,2,f,g-l,ntot-g,pval)
           if (pval.lt. 0.001)  pval =  0.001

           writefiunit, ' (//) ')
           if(ipd).lt.l)  then
                                m              Actable difference for',

          else
        write(iunit,'(a,/,a,fl5.6)»)  ' Minumum detectable difference for'
       &    t-tests with Bonferroni adjustment « • mdd
          endif
                                       371

-------
'.4   Listing of  Computer Program for Dunnett's Procedure  (Continued).
            if(inum.gt.l) write(iunit,'(a,f15.6,a)»)
           1 This corresponds to  a difference of  ',mddl,' in original units
            write(iunit,'(a,f8.2,a//)')  '  This difference corresponds to  ',
   100
&  pctrl, '  percent of control'

   if(ieqn.eq.l)  then
  write(iunit,*j
  write(iunit,*)
  write(iunit,*)
  writefiunit,*)
  writefiunit,*)
  write(iunit,*)
  write(iunit,*)
  writefiunit,*)
  write(iunit,*
   end if
                          * Note - the above value for the minimum       *
                          * detectable difference is approximate as      *
                          * the sample sizes are not the same for all of *
                          * the groups.                                  *
                                                                         *
           write(iunit,'(a,f!6.6,a,i2,a/J')
             r Between groups sum of squares =l,ssb,' with
               degrees of freedom.
                                       p
                                    ', pval

                             ssw/idf2,  • with  ',idf2,
write(iunit,'(a,f6.3,/)'
write(iunit,'(a,fl6.6,a,
  '  Error mean square -
  *  degrees of freedom,'
if(p.gt.l.dO)  goto 100
if(p,It.0.001)  then
  p  «= 0.001
  write(iunit,*(a,f6.3,/)•)
  1  Bartletf'e  test p-value for equality of variances <=
  else
  write(iunit,'(a,f6.3,/J r)
  '  Bartletf's  test p-value for equality of variances = '
  endif
              return
  if(p.gt.0.01)
 write(iunit,*)
 write(iunit,*)
 write(iunit,*j
 write(iunit,*)
 write(iunit,*)
 writefiunit,*)
 write(iunit,*)
 write(iunit,*)
  return
 write(iunit,*)
 write(iunit,*)
 write(iunit,*)
 write(iunit,*)
 write(iunit,*)
 write(iunit,*)
 write(iunit,*)
  return
  end
                          *                                               *
                          *  Warning -  the test  for equality  of  variances  *
                          *  is significant  (p  less  than  O.bl).  The
                          *  results  of  this  analysis  should be  inter-
                          *  preted with caution.
                          *
                         '*                                              *
                         '* Warning - the test  for equality of variances *
                         1 * could not be computed as 1 or mere of the    *
                         '* variances is zero.                           *
                         '*                                              *
                                        372

-------
I
12.4  Listing  of Computer  Program for Dunnett's  Procedure  (Continued).
     $storage:2
           REAL*8 FUNCTION TINV(P,NDF)
           IMPLICIT REAL*8(A-H,0-Z)
           DF=NDF*1.DO                                                   :
           Z-GAUINV(P)
           T=Z
           T=T+(Z**3+Z)/(4.DO* DF)
           T=T4(5.DO*Z**5+16.DO*Z**3+3.DO*2)/(96.DO* DF**2)
           T=T+(3.DO*Z**7+19.DO*Z**5+17.DO*Z**3-15.DO*Z)/(384.DO*  DF**3)
           T=T+(79.DO*Z**9 + 776.DO*Z**7-t-1482.DO*Z**5-1920.DO*Z**3-945.DO*Z)/
          $(92160.DO* DF**4)
           TINV=T
           RETURN
           END
           REAL*8 FUNCTION GAUINV(P)
           IMPLICIT REAL*8(A-H,0-Z)
           D=P
           IF(D.GT..5DO) D=1,-D
           T2=DLOG(1./(D*D)J
           T=DSQRT(T2)
           GAUINV=T-(2.515517DO-t-0.802853DO*T+0.010328DO*T2)/
          *         (1.0D04l.432788DO*T+0.189269DO*T2+.001308DO*T*T2)
           IF(P.LT..5DO) GAUINV^-GAUINV
           RETURN
           END
                                          373

-------
2.4  Listing of Computer Program  for Dunnett's  Procedure  (Continued).
      $ include:'speedy.fil'
      c                                                  0  ... 79
      c   cursor  positioning,  goto (irow,icol)            :
      c                                                 24
            subroutine gotorc (irow,icol)
            character*! dummy

            if  (irow.lt.0.or.irow.gt.24)  irow =  0
            if  (icol.lt.0,or.icol.gt,79)  icol =  0
            read (dummy,'(Ix)')
            call locate (0,0,ier)
            write <*,'(\)')
            call locate (irow,icol,ier)
            read (dummy,'(Ix)')
            return
            end
      c  clears the screen,  returns  to  text mode  if not  already  in text mode,
      c  and puts cursor in  upper left  corner

            subroutine clrscr

            call qrmode (im,i)
            if (im.ne.2)  call qsmode(2)
            call gclear (0,7)
            call gotorc (0,0)
            write (*,'(\)')
            return
            end


      c  clears to end of line from  current cursor position, does not move
      c  cursor

            subroutine clreol

            call gcpos (icol,irow)                                      .
            if (icol.eg.79)  then
              if (irow.ne.24)  write  (*,'(al)')  *  '
            else
              call qstext (32,0,7,(79-icol))
            end if
            return
            end
     c  gotorc + clreol

           subroutine goclrc  (irow,icol)

           call gotorc  (irow,icol)
           call clreol
           return
           end
                                         374

-------
2.4  Listing of Computer  Program for  Dunnett's Procedure {Continued}
     c  converts  a  character  (ascii value) to upper case  (if lower case)

           character*!  function upchar  (ch)
           character*!  c,ch
           equivalence  (ich,c)
           c  =  ch
           if (ich.ge.97.and.ich.le.122)  ich
           upchar  = c
           return
           end
ich - 32
     c  converts  a  fortran string  of  length  len to uppercase

           subroutine upstr (str.len)

           character*! str(l)
           character*! upchar
           do 10  i=l,len
     10    str(i) - upchar(str(i)}
           return
           end


     c  back up (move left)  one text  position  on  the  screen

           subroutine backup

           call qcpos (icol,irow)  .
           call qcmov ((icol-1),irow)
           return
           end


     c  pauses for  idelay seconds

           subroutine delay (idelay)

           call qtime (ihr, iinin, isec, ihun)
     10    call qtime (ihr,imin,nsec,ihun)
           if (isec.gt.nsec) then
             iexp = 60 - isec  + nsec
           else
             iexp = nsec - isec
           endif
           if (iexp.gt,idelay)  return
           goto 10
           return
           end


     c  get a string from the  keyboard,  echo it to the  screen, but handle
                                          375

-------
12.4  Listing of Computer Program  for Dunnett's  Procedure (Continued).
       c  each character separately,  so there is no scrolling,  etc.
       c  if user hits escape,  str(l)  = ESC (char(27)),  then return.
       c  if string is of length zero, str(l)  » char(13J.
       c  maximum length of string accepted is len.

             subroutine getstr  (str,len)
             character*! str(1),gstr(80)
             equivalence (gstr(l),ifirst)

             ilen « 0
       10     call qinkey (iext.key)
             if ((iext.eq.l).and.((key.eq.13).or.(key.eq.27)))  goto 100
             if (iext.eq..Land.key.ge. 32. and.Key. le.126.and.ilen.lt. len) then
               write (*,'(al,\)') key
               ilen = ilen + l
               gstr(ilen)  = key
             endif
             if (iext.eq.Land.key.eq.8.and.ilen-.gt.O) then
               call backup
               write (*,'(al,\) ')  ' '
               call backup
               ilen = ilen -  l
             end if
             goto 10

       100    if (key.eq.27) then
               ifirst =  27
               ilen = 1
             else
               do 105 k=ilen+l,len
       105      gstr(k) =  •  *
             end if
             if (ilen.eq.O) ifirst = 13

             do lio  k=l,len
       110    str(k)  =  gstr(k)
             return
             end


      c  this  routine gets a password for a protected data set.
      c  it works about the same as getstr, but does not echo the input
      c  characters to the screen,   maximum length is 10 characters.

            subroutine gpwrd (gstr)
            character*! gstr(10)
      10
ilen = o
call qinkey (iext,key)
if (iext.eq,0) goto 10
if (key.eq.27) goto 27
if (key.eq.13) goto 13
if (key.ge.32.and.key.le.126.and.ilen.lt.10)  then
  write (*,'(a\)')  *  '
  ilen « ilen + 1
                                          376

-------
1.4   Listing of  Computer Program for Dunnett's  Procedure  (Continued)
   13
   27
     gstr(ilen)  - key
    end if
    if (key.eq.S.and.ilen.gt.O) then
     call backup
     ilen =  ilen - 1
    endif
    goto 10
    if (ilen.eq.O) goto 27
    call pad  (gstr,ilen,10)
    return
    gstr(l) - char(key)
    return
    end
   c  this routine decodes an encrypted password stored in the header
   c  file and returns the original ascii string.
         subroutine dpwrd (pw)
         character*! pw(10), n(10)
call copy (pw,
pw(l)
pw(2)
pw(3)
pw(4)
pw(5)
pw(6)
pw(7)
pw(8)
pw(9)
pw(10)
BE
*
=
=
as
t*
*
*
=
=
char
char
char
char
char
char
char
char
char
char
n,10)
(ichar(n(6))
(ichar (n (3) )
(ichar(n(8) )
(ichar(n(4) )
(ichar(n(10))
(ichar(n(7) )
(ichar(n(5) )
(ichar(n(2) )
(ichar (n(9))
(ichar(n(l))
         return
         end
                                       2 +
                                       2 +
                                       2 +'
                                       2 +
                                       2 +
                                       2 +
                                       2 +
                                       2 +
                                       2 +
                                       2 4
                                     6)
                                     3)
                                     8)
                                     5)
                                     9)
                                     4)
                                     2)
                                     1)

                                     0)
   10
   20
finds length of a string (str)  of max length mien (position  of  last
non-blank character)

   function length (str(mlen)
   character*! str(mien)

   k - mien 4 1
   k = k - l
   if (k.eq.O) goto 20
   if (str(k).eq.•  ')  goto 10
   length - k
   return
   end
   c  centers a string (str)  in a  field of len characters, padded on both
   c  sides  by blanXs.   strips  away  leading blanks, then pads it back out,
                                        377

-------
2.4
Listing  of Computer Program for Dunnett's Procedure  (Continued)
    10



    30

    40
     subroutine  center  (str,len)
     character*! str(l)

     if  (length(str,len).eq.O) return
     kl  = 0
     kl  - kl + 1
     if  (str(kl).eg.' '} goto 5
     k2  - length (str,len}
     ilen - k2 - kl + 1
     do  10 k-kl,k2
     str(k-kl-H)  * str(k)
     call pad (str,ilen,len)
     imov * (len-ilen)/2
     do  30 k=ilen,l,-l
     strfk+imov)  = str(k)
     do  40 k=l,imov
     str(k)  - '   *
     return
     end
   10
 left justifies a string, pads with blanks to the right

    subroutine leftj (str,len)
    character*! str(l)

    if (length(str,len).eq.O)  return
    kl = 0
    kl - kl + l
    if (str(kl).eq.' »)  goto 5
    k2 = length (str,len)
    ilen - k2 - kl + l
    do 10 k=kl,k2
    str(k-kl-t-i) = str(k)
    call pad (str,ilen,len)
    return
    end
   10
      right justifies a string of length len, pads with blanks at left

         subroutine rightj (str,len)
         character*! str(l)
    if  (length(str,len).eq.O) return
    kl  «  0
    kl  =  kl + 1
    if  (str(kl).eq.'  ') goto 5
    k2  «=  length  (str,len)
    ilen  « k2 -  kl  +  l
    do  10 k«k2,kl,-l
    str(k-k2+len) « str(k)
    call  pad (str,0,len-ilen)
    return
    end
                                        378

-------
               ;

   Listing  of Computer  Program for Dunnett's Procedure  (Continued)
  c  strips the dollar sign  (col 1) from a fortran string of length len
  c  pads with a blank at the right.                         *««gtn ien,

       subroutine stripp (str,len)
       character*! str(l)

       if (str(l).eq.'$') then
         do 10 k=l,len-l
  10        str(k) = str(k+!)
         str(len) = ' '
       end if
       return
       end


 c  copies LEN characters from fortran string FROM to fortran  string TO

       subroutine copy (from,to,len)
       character*! from(len),to(len)

       do 5  k=l,len
 5     to(k)  » from(k)
       return
       end


 c  pads  fortran  string  of length  LEN  to length NEWLEN with blanks

       subroutine  pad  (str, len,newlen)
       character*! str(newlen)

       do  5 k=len-n,newlen
 5     str(k) m '  "
       return
       end
c  forces user to press a function key between Fl and Fn.
c  ESC is considered equivalent to Pi.
10
20
function ifnkey  (n)

call qinkey  (i,k)
if  (i.eq.o.and.k.ge.59.and.k.le.(584n)}  goto 20
if  (i.eq.l.and.k.eq.27) then
  ifnkey = 1
  return
end if
goto 10
ifnkey - k - 58
return
end
                                     379

-------
2.4  Listing of Computer Program  for Dunnett's Procedure (Continue^)
     c   tests whether  the  first  len characters of two strings are
     c   identical.

           logical  function  equal  (sl,s2,len)
           character*! sl(l),s2(l)

           equal  -  .true.
           do 5 k=l,len
     5      if  (sl(k).ne.s2(k)) goto 10
           return
     10     equal  »  .false.
           return
           end
    c  tests  for  "same1* col name - upcase, stripp

           function  same (a,b,len)
           character*! a(len),b(len),al(80),a2(80)
           logical same,equal

           call copy (a,al,len)
           call copy (b,a2,len)
           call upstr (al,len)
           call upstr (a2,len)
           call stripp (al,len)
           call stripp (a2,len)
           if  (equal(al,a2,lenj) then
            sane =  .true.
           else
            same =  .false.
           endif
           return
           end
       tests for upper-case equality of strings - capitalizes,  then tests,
          function uequal (a,b,len)
          character*! a(len),b(len),al(80),a2(80)
          logical uequal,equal

          call copy (a,al,len)
          call copy (b,a2,len)
          call upstr  (al.len)
          call upstr  (a2,len)
          if (equal(al,a2,len)J then
            uequal = .true.
          else
            uequal « .false.
          endif
          return
          end

    c  prompts user to press any key to continue.
    c  at the bottom of the screen.

          subroutine retcon
message is centered
          call goclrc (23,27)
          write (*,'(a\)')  'press any key to continue'
          call gotorc (23,27)
          call qinkey (ji,ji)
          return
          end
                                         380

-------
                                  APPENDIX D

                              BONFERRONI'S T-TEST


 1.   Bonferroni's  T-test  Is  used as an alternative to Dunnett's Procedure
[when the  number of  replicates  is  not the  same for all concentrations.  This
[test sets an  upper  bound of alpha on the  overall error rate, in contrast to
•Dunnett's Procedure, for which the overall error rate is fixed at alpha.
[Thus,  Dunnett's Procedure  is a more powerful test.

  .   Bonferroni's  T-test  is  based  on the same assumptions of normality of
 distribution  and  homogeneity of variance  as Dunnett's Procedure (See,
[Appendix  B for testing these assumptions), and,  like Dunnett's Procedure,
 uses a pooled estimate of  the  variance, which is equal to the error value
 calculated in an  analysis  of variance.

 3.   An example of the use  of Bonferroni's T-test is provided below.  The
 data used in  the  example are the  same as  in Appendix C, except that the
 third  replicate from the 32% effluent treatment  is presumed to have been
 lost.   Thus,  Dunnett's Procedure  cannot be used.  The weight data are
 presented in  Table  D.I.
         TABLE  D.I.  SHEEPSHEAD  LARVAL GROWTH  DATA  (WEIGHT IN MG)
                    USED  FOR  BONFERRQNI'S TEST
Effluent     i

Cone {*)
Replicate Test Vessel

1           2        3
                                                     Total
Mean
Control
1.0
3.2
10.0
32.0
1
2
3
4
5
1.017
1.157
0.998
0.873
0.715
0.745
0.914
0.793
0.935
0.907
0.862
0.992
1.021
0.839
(Data lost)
2.624
3.063
2.812
2.647
1.622
0.875
1.021
0.937
0.882
0.811
 ^Prepared  by Ron  Freyberg,  Florence  Kessler,  John Menkedick and Larry
 Wymer,  Computer Sciences  Corporation,  26 W. Martin Luther King Drive,
 Cincinnati,  Ohio  45268; Phone  513-569-7968.
                                     381

-------
        *.,in ?? estimate of the P°oled variance  is  to  construct  an
        table including all sums  of squares,  using the  following formulas:

  Total  Sum of Squares:    SST  = I  Y? .  -  G2/N
Between Sum of Squares: SSB = Z T?/n .  -  G2
                              i  7  7
Within Sum of Squares:  SSW = SST - SSB
                                             /N
    Where:  G = The grand total  of all  sample observations;  G
           N = The total sample size;  N = E n.
                                                                E T.
                                                                1
           n. - The number of replicates for concentration "i".

           Ti = The total of the replicate measurements for concentration "i"

          Yij = The ^th observation for concentration "i".

 3.2  Calculations:

 Total Sum of Squares:    SST = I  Y?.  -  G2/N
                                i -i   J
                              - 11.832 -

                              = 0.188
                                        12.768'
                                          14
 Between Sum of Squares:   SSB  =  £  T?/n.  -   G2/N

                              =  11.709  -  (12.768}2/14
                              =  0.064

[Within  Sum  of  Squares:    SSW  =  SST  - SSB
                              =  0.188 - 0,064
                              =  0.124
                                       382

-------
3.3 Prepare the ANOVA table as follows:
                      TABLE D.2.   GENERALIZED ANOVA TABLE
Source
  DF      Sum of
          Squares {SS)
                                            Mean Square (MS)
                                              (SS/DF)   .
Between    b - 1
Within     N - b
                 SSB
                            SSW
                                                 = SSB/(b-l)
                                      = SSW/(N-b)
Total
N - 1
                            SST
*b = Number of different concentrations, including the control.
3.4  The completed ANOVA table for this data is provided below;
     TABLE D.3. COMPLETED ANOVA TABLE FOR BONFERRONI'S TEST
Source
   DF
                             SS
Mean Square
Between    5-1=4      0.064

Within    14 - 5 =  9      0.124
                                  0.016

                                  0.014
Total
  13
                           0.188
                                      383

-------
 3.5  To perform the  individual  comparisons, calculate the  t  statistic for
 each  concentration and  control  combination, as  follows:


                                    (yi  -V
             [S
                               w
                                         + (l/n.J]
    Where:
            w


           nl


           ni
=  Mean for each concentration


=  Mean for the control


=  Square root of the within mean square


=  Number of replicates in the control.


=  Number of replicates for concentration "i"
3.6  Table D.4  includes the calculated t values for each concentration and
control combination.
                       TAB.E 3.4.   CALCULATED  T  VALUES.
Effluent
Concentration i
1.0 2
3.2 3
10.0 4
32.0 5

11
- 1.511
- 0.642
- 0.072
- 0.592
                                     384

-------
3.7  Since the purpose of the test is only to detect a decrease in growth
from the control, a one-sided test is appropriate.  The critical  value for
the one-sided comparison (2.686), with an overall  alpha level  of  0.05, nine
degrees of freedom and four concentrations excluding the control, was
obtained from Table D.5.  Comparing each of the calculated t values in
Table D.4 with the critical value, no decreases in growth from the control
were detected.  Thus the NOEC is 32.0%.
                                    385

-------
TABLE 0.5. CRITICAL VALUES FOR BONFERRGNI'S "T"
      P = 0.05 CRITICAL LEVEL, ONE TAILED
i
D.F.
2
f £

I 5
i $
i i
1 !
i ,2
? 10
i u
£ 12
1 1!
I JJ

f 16
? ,17
1 8
L I9
20
21
23
24
25
?*
2 7
28
29
30
31
22
33
34
35
26
3?
39
39
40
50
60
70
£0
SO
103
110
120
IMF.
D.F. =
K =
K
6
2
- 1 K = 2
.314 12.707
.920 4,303
2. 354
2.132
4
i
i
:
;
i
i
]
]
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
t
i


.016
.94*
.895
. 660
.834
.6)13
. 796
.783
.771
. 762
. 754
.746
. 740
.735
.730
. 725
.721
. 718
. 714
.711
.709 .
. 706
. 704
. 702
. 700
.6*8
.696
.694
.693
.691
.690
.689
.688
.686
.685
.684
.676
.671
.66?
.665
.662
.661
.659
658
.645
Degrees
Number
3.183
2.777
2.571
2.447
2.365
2.307
2.263
2.229
2.201
2. 179
2.161
2. 145
2. 132
2.120
2. 110
2. 101
2.394
2.086
2.080
2.074
2.069
2.064
2.060
2.056
2.052
2.049
2.046
2.043
2.040
2.037
2.035
2.033
2.031
2.029
2.027
2.025
2.023
2.022
2.009
2.001
1.995
1.991
1.967
1 . 5 64
1.9&2
1.980
1.960
of
K = 3
19.002
5.340
3. 741
3.187
2.912
2. 750
2,642
2.567
2. 510
2.4(6
2.432
2.404
2,380
2.360
2.343
2.329
2.316
2. 3C5
2.295
2.206
2.278
2.271
2.264
2.258
2.253
2.246
2.243
2.239
2.235
2.231
2.228
2.224
2. 221
2.219
2.216
2.213
2.211
2.209
2,207
2.205
2.189
2. 179
2.171
2.166
2. 162
2,158
2.156
2.153
2.129
freedom
K = 4
25.452
6.206
4. 177
3.496
3.164
2.969
2. 842
2.752
2.686
2.634
2.594
2.561
2. 533
2.513
2.490
2.473
2.459
2.446
2.434
2.424
2.414
2.406
2 .398
2,391
2.385
2.579
2.374
2.369
2.364
2. 360
2.356
2.352
2. 349
2.346
2.342
2. 340
2.33?
2.334
2.332
2.329
2.311
2. 3UO
2. 291
2.285
2.260
2. 276
2.273
2.270
2.242
for MSE
of concentrations to
K = 5
31.821
6.965
4.541
3. 747
3.365
3. 143
2.998
2.697
2.822
2. 764
2.719
2.661
2.651
2.625
2.603
2.584
2.567
2.553
2.540
2.526
2.518
2. 5C9
2.500
2.493
2.486
2.479
2.473
2. 468
2,463
2.458
2. 453
2.449
2.445
2. 442
2.436
2.435
2, 432
2.429
2.426
2. 424
2.404
2.391
2. 361
£•3 ?^
2.369
2.365
2. 361
2. 353
2. 327
(Mean
K = 6
38. 189
7.649
4.657
J.961
3.535
3.288
3.128
J.016
2.934
2.871
2.821
2.780
2. 746
2.718
2.6^4
2.674
2.655
2.64C
2.626
2.613

2.'592
2.583
2.574
2.566
2.559
2.553
2.547
2.541
2 .536
2. 531
2.527
2.523
2. 519
2.515
2.512
2.508
2.505
2.502
2.499
2.476
2.463
2.453

2^440
i. '35

21*29
2.354
Square
be compared
K = 7
44.556
8. 277
5.138
4.148
3.681
3.412
3.239
3.118
3 .029
2.^61
2.907
2.S63
2. 827
2.797
2.771
2. 749
2. 729
2.712
2.697
2.684
2.672
2. 661
2.651
2.642
2.634
2.627
2.620
2.613
2.607
2.602
2.597
2.592
2.587
2.583
2.579
2.575
2.572
2.560
2.565
2.562
2.539
2.524
2.513
2. 5-C5
2.495
i .494
2 . *• 9 D
2.4g7
2.453
Error)
to the
X = 8
50.924
8. 661
5.392
4. 315
3. 811
3.522
3.336
3. 206
3.111
3.039
2. 981
2 .935
2. 897
2.844
2 .33 7
2.814
2. 793
2.775
2. 759
2. 745
2.732
2. 721
2. 710
2.7C 1
2.692
2. 464
2.677
2.670
2. 664
2.658
2 .652
2. 647
2.643
2.633
2.C34
2.63D
2 .626
2. 623
2.619
2. 616
. 592
.576
. .564

.5*9
C44
.* \ 4 0
2.536
2 .458
K = 9
57.290
9.406
5.626
4.466
3.927
3.619
3.422
3. 285
3. 185
3.108
3. 047
2.998
2.958
2.924
2.C95
2.871
2.849
2.830
2.61 3
2. 798
2. 785
2.773
2. 762
2.752
2.743
2. 734
2. 727
2.720
2.713
2.707
2.701
2. 696
2. 691
2.686
2. 682
2. 678
2.674
Z. 67C
2. 667
2.663
2. 638
2.621
2.609
2. 6CC
2. 593
2.588
2. 583
2. 580
2.540
K = 10
63.657
9. 925
5.841
4 .605
4. 033
3.708
3 .500
3.356
3.250
3.170
3. 106
3.055
3.013
2.977
2.94?
2 .921
2 . 899
2 . 879
2 .861
2. 846
2. 832
2 .819
2. 808
2 . 797
2.788
2.779
2.771
2.764
2.757
2. 750
2.745
2 . 739
2.734
2 .729
2. 724
2. 720
2.716
2.712
2.708
2.705
2.678
2. 661
2.648
2. 639
2. 632
2 .626
2. 622
2.618
2.576
from ANQVA.
control .


                      386

-------
                                   APPENDIX E

                        •   STEEL'S MANY-ONE RANK TESll

11.  Steel's Many-One Rank Test is a nonparametric test for comparing
^treatments with a control.  This test is an alternative to the Dunnett's
  Procedure, and may be applied to the data when the normality assumption has
  not been met.  Steel's Test requires equal variances across the treatments
 ;and the control, but it is thought to be fairly insensitive to deviations
|from this condition (Steel, 1959).  The tables for Steel's Test require an
 [equal number of replicates at each concentration.  If this is not the case,
 .use Wilcoxon's Rank Sum Test, with Bonferroni's adjustment (See Appendix F).

 12.  For an analysis using Steel's Test, for each control and concentration
 [combination, combine the data and arrange the observations in order of size
 'from smallest to largest.  Assign the ranks to the ordered observations (1
  to the smallest, 2 to the next smallest, etc.).  If ties occur in the
  ranking, assign the average rank to the observation.  (Extensive ties would
  invalidate this procedure.)  The sum of the ranks within each concentration
  and within the control is then calculated.  To determine if the response in
  a concentration is significantly different from the response in the control,
  the minimum rank sum for each concentration and control  combination is
  compared to the critical value in Table E.5.   In this table, k equals the
  number of treatments excluding the control and n equals  the number of
  replicates for each concentration and the control.

  3.  An example of the use of this test is provided below.   The test employs
  survival data from a mysid 7-day, chronic test.  The data  are listed in
 Table E.I.  Throughout the test,  the control  data are taken from the site
 water control.  Since there is 0% survival for all eight replicates for the
  32% concentration,  it is not included in this analysis and is considered a
  qualitative mortality effect.

 4.  For each control and concentration combination,  combine the data and
 arrange the observations in order of size from smallest  to largest.  Assign
 the ranks (1,2,3,  ...  16) to the ordered observations  (]  to the smallest,
 2 to the next smallest,  etc.).  If ties occur in the ranking, assign the
 average rank to the observation.

 5.  An example of  assigning ranks to the combined data for  the control  and
 0.32% effluent concentration is  given in Table E.2.   This  ranking procedure
 is repeated for each control and  concentration combination.   The complete
 set of rankings is  listed  in Table E.3.   The  ranks are then  summed for  each
 effluent concentration,  as shown  in  Table E.4.
 ^Prepared by Ron Freyberg,  Florence Kessler,  John  Menkedick and  Larry
 Wymer,  Computer Sciences Corporation,  26 W. Martin Luther King Drive,
 Cincinnati,  Ohio 45268;  Phone 513-569-7968)

                                      387

-------
jj6.   For this set of data,  we wish to determine  if  the  survival  in any of the
jef fluent concentrations is significantly  lower  than  the  survival of  the
|control  organisms.   If this occurs,  the rank  sum at  that concentration would
fbe significantly lower than the rank sum  of the control.   Thus, we are only
Iconcerned with comparing the rank sums for the  survival  at each of the
Ivarious  effluent concentrations with some "minimum"  or critical rank sum, at
[or below which the  survival  would be considered to be  significantly  lower
fthan the control.   At a probability  level of  0.05, the critical rank sum in
   test  with  four, concentrations and  eight replicates per concentration,
|is  47 (see Table F.4).

     Of  the rank  sums  in Table E.4, none are less than  47.   Therefore, due to
fthe qualitative  effect at  the 32% effluent concentration,  the NOEC is 10%
[effluent and the LOEC is 32% effluent.
                                     388

-------
TABLE E.I. EXAMPLE OF STEEL'S MANY-ONE RANK TEST-
           SURVIVAL DATA FOR MYSID 7-DAY CHRONIC TEST
Effluent
Concentration



Control
(Site Hater)






Control
(Brine «
Dilution water)





0.32?







1.0?







3.2%







10.0*







32. 01




Replicate
Chamber
1
2
3
4
5
6
7
8
1
2
3
4
5
6
7
8
1
2
3
4
5
6
7
8
1
2
3
4
5
6
7
8
1
2
3
4
5
6
7 •
8
1
2
3
4
5
6
7
8
1
2
3
4
5
6
7
8
Number of
My s Ids at
Start of Test
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
• 5
5
5
5
5
5
5
5
5
5
5
Number of
Live Mysids
at End of Test
4
4
5
4
5
4
4 '
5 ;
3
5
3
3
4
4
3
3
4
4
4
5
4
4
5
3
3
4
5
4
4
4
5
5
5
4
5
3
5
4
4
3
5
5
5
5
3
5
4
4
0
0
0
0
0
0
0
0
                       389

-------
TABLE E.2. EXAMPLE OF STEEL'S MANY-ONE RANK TEST: ASSIGNING
           RANKS TO THE CONTROL AND 0.32% EFFLUENT CONCENTRATIONS
  Rank      Number of Live
                Mysids
                  Control or 0.32% Effluent
  1
  6.5
  6.5
  6.5
  6.5
  6.5
  6,5
  6.5
  6.5
  6.5
  6.5
14
14
14
14
14
 3
 4
 4
 4
 4
 4
 4
 4
 4
 4
 4
 5
 5
 5
5
5
 0.32%
 Control
 Control
 Control
 Control
 Control
 0.32%
 0.32%
 0.32%
 0.32%
 0.32%
 Control
Control
Control
0.32%
0.32%
                       TABLE  E.3.   TABLE  OF RANKS
1 It Replicate
ml Lnamber
as I
ill 4
H2 4
H3 5
H4 4
I6 4
1 l
Control 1
" i —
{6.5,6,6.
(6.5,6,6.
(14,13.5,
(6.5,6,6.
£14,13.5,
(6.5,6,6.
(6.5,6,6.
(14,13.5,
~ • 	
5,5}
5,5)
13.5,12.5}
5,5}
13.5,12.5)
5,5)
5,5)
13.5.12.5}
U
4
4
4
5
4
4
5
3
Effluent Concentrating (%}
.32
^ ^-^— *™^
(6.5)
(6.5)
(6.5)
(14)
(6.5)
(6.5)
(14)
(1)
1
3
4
5
4
4
4
5
5
.0
, — __
(1)
(6)
(13.5)
(6)
(6)
(6)
(13.5)
(13.5)
3
5
4
5
3
5
4
4
3
.2
— 	 — •
(13.5)
(6.5)
(13.5)
(1.5)
(13.5)
(6.5)
(6.5}
(1.5)
10.0
* i .
5 (12.5)
5 (12.5)
5 (12.5)
5 (12.5)
3 (1)
5 (12.5)
4 (5)
4 (5)
                         °rder  °f  the  Concentration with which  they
                               390

-------
                        TABLE E.4.  RANK SUMS
              Effluent
           Concentration
Rank Sum
                0.32
                1.00
                3.2
               10.0
   61.5
   65.5
   63.0
   73.5
TABLE E.5.  SIGNIFICANT VALUES OF RANK SUMS:  JOINT CONFIDENCE
           COEFFICIENTS OF 0.95 (UPPER)  and  0.99 (LOWER)  FOR
           ONE-SIDED ALTERNATIVES
n
4
5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

k =
2
11
18
15
27
23
37
32
49
43
63
56
79
71
97
87
116
105
138
125
161
147
186
170
213
196
241
223
272
252
304
282
339
315
number
3
10
17
—
26
22
36
31
48
42
62
55
77
69
95
85
114
103
135
123
158
144
182
167
209
192
237
219
267
248
299
278
333
310
of treatments
4 5
10
17
—
25
21
35
30
47
41
61
54
76
68
93
84
112
102
133
121
155
142
180
165
206
190
234
217
264
245
296
275
330
307
10
16
• —
25
21
35
30
46
40
60
53
75
67
92
83
111
100
132
120
154
141
178
164
204
188
232
215
262
243
294
273
327
305
(excluding
6 7
10
16
_
24
_
34
29
46
40
59
52
74
66
91
82
110
99
130
119
153
140
177
162
203
187
231
213
260
241
292
271
325
303
-
16
_
24
_
34
29
' 45
40
59
52
74
66
90
81
109
99
129
118
152
139
176
161
201
186
229
212
259
240
290
270
323
301
control )
8 9
-
16
_
24
_
33
29
45
39
58
51
73
65
90
81
108
98
129
117
151
138
175
160
200
185
228
211
257
239
288
268
322
300
_ ' ' -
15

23

33
29
44
39
58
51
72
65
89
80
108
98
128
117
150
137
174
160
199
184
227
210
256
238
287
267
320
299
                1959.
                           391

-------
                                    APPENDIX F

                              WILCOXON RANK SUM TEST


  1. Wilcoxon's Rank Sum Test is a non-parametric test,  to be used  as  an

  nl 'thP ^ t0t Ste6 'S Many-°ne Rank "est whe" the n^ber of replica?es  are
  not the same at each concentration.   A Bonferroni's adjustment  of the
      W<0T°r rate f°J comPa^°n °f each concentrat  o   vTtVcon  rol  is
             Ma" Ter b°Und of a]Pha  on the overall  error rate,  in contrast
          Ih "an£0n',Ra?k T«t. f°r "hich the  overall error  rate Is fixed Jt
          Thus, Steel's  Test is a more powerful  test.
      e      in
                     hsev    r^ice
                                            wl'th reproduction data from the

 Observation
L,n°I tea^?^n!ra.t.!°,n.a!;d "zlrfrrs^ne^to^ar"6  ^  "** •  ^

                          in rank occur,  assign  the  average  rank  to the


effl^f^L^fL5:9"1',"!/?^?,10  the  combined data for the control and
 5.  For this set of data, we wish to determine if the fertility  in  anv of

      ?s thCrra%npraJh°nS VS S1'9"jfl'Cantly lo«er than  in the  contro  .  When
 contro! i5?i K!'- thVankJUtn1for that """ntration compared to the
 control W111 be significantly lower than the  rank sum given by the  averaae
frank over both concentrations times the number of replicates at  that test

         f Teacher ^^r^  WUh C0m"ar1n   thfrank sum  V  S
 crit     rlnt, Inn,  VhS ft1"6^.00"06"^3110"5  W1'th  some  '"Inimum" or
[critical rank sum, at or below which  the fertility would be considered to he
 sigmncantly  ower than the control.   At a probability level of 0 OS  the
 critical rank in a test with four  concentrations  and  seven repllcates'in the
|control is 44 for those concentrations  with eight  replicates  and 34 for
jthose  concentrations with  seven replicates (sel Table F?5. fSr R - 4).

rn  t  °f  th!! ^ank  s™s 1n Tab]e  F'4>  only ^e 10% effluent concentration  does
fnot  exceed _ its  critical  value  of 44.  Therefore, the LOEC for ?he test  on
fertilTty  1S  10% effluent, and the  NOEC  is 3.2% effluent
  L P  r  by Ron Freyberg, Florence Kessler, John Menkedick  and Larry
Wymer, Computer Sciences Corporation, 26 W. Martin Luther Kina Drive
Cincinnati, Ohio 45268; Phone 513-569-7968)                  9      '
                                     392

-------
TABLE F.I. EXAMPLE OF WILCOXON'S RANK SUM TEST-
           FECUNDITY DATA FOR MYSID 7-DAY CHRONIC TEST
Effluent
Concentration


Control
(Site Water)





Control
(Srine &
Dilution water)




0.32?






1 fW
. ire-






3.2$






10.0?







32.0%




Replicate
Chamber
1
2
3
4
5
6
7
8
1
2
3
4
5
6
_ 7
8
1
2
3
4
5
6
7
8
1

3
4.
5
6
7
8
1
2
3
4
5
6
7
8
1
2
3
4
5
6
7
8
1
2
•3
4
5
6
7
8
Number of
Hysids at
Start of Test
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5 -'•; '•
5
5
5
5
5 ..-.
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
Number of Proportion
Live Mysids • of Females
at End of Test With Eggs
4 0.50
4
5 0.75
4 0.67
5 0.67
4 0.50
4 1.00
5 1.00
3 1.00
5 1.00
3 1.00
3 1.00
4 1.00
4 0.50
3 0.50
3 0.50
4 1.00
4 0.50
4 0.67
5 1.00
4 0.50
4 1.00
5 1.00
3 0.00
3 0.50
4 0.00
5 0.75
« 1 . 00
4 l . 00
.. '* 1.00
' -' 5 0.67
i! 5. 0.67
5 0.33
4 0.50
5 1.00
3
5 1.00
4 0.00
4 0.33
3 0.50
5 0.00
5 0.50
5 0.33
5 0.00
3 0,50
5 0.00
4 0.50
4 Q.50
. •• 0
0
0
0
0
0
0
0
                    393

-------
TABLE F.2. EXAMPLE OF WILCOXON'S RANK SUM TEST: ASSIGNING
           RANKS TO THE CONTROL AND EFFLUENT CONCENTRATIONS
 Rank      Proportion of
           Females W/Eggs
Site Water Control
or 32% Effluent
1
3.5
3.5
3.5
3.5
7
7
7
9
12.5
12.5
12.5
12.5
12.5
12.5
0.00
0.50
0.50
0.50
0.50
0.67
0.67
0.67
0.75
1.00
1.00
1.00
1.00
1.00
1.00
0.32%
Control
Control
0.32%
0.32%
Control
Control
0.32%
Control
Control
Control
0.32%
0.32%
0.32%
0.32%
I TABLE F
I Site Water
j 1 Rep Proper- Control Rank
i i
1
I
I ;

i |
|| :
•' i
1


1
2
3
4
5
6
7
8
tion
0.50 (3.5,3,5.5,7,5)
- - - -
0.75 (9,9.5,10,13)
0.67 (7,6.5,8.5,11.5)
0.67 (7,6.5,8.5,11.5)
0.50 (3.5,3,5.5,7.5)
1.00 (12.5,13,12.5,14.
1.00 (12.5,13,12.5,12.
aControl ranks are given in
.3. TABLE OF
Effluent
0.32
1
0
0
1
0
1
5) 1
5) 0
the
.00
.50
.67
,00
.50
.00
.00
.00
(12.5)
(3.5)
(7)
(12.5)
(3.5)
(12.5)
(12.5
(1)
order of the
RANKS 1




Concentration (%)
.,1.0
0.50
0.00
0.75
1.00
1.00
1.00
0.67
0.67
(3)
(1)
(9.5)
(13)
(13)
(13)
(6.5)
(6.5)
3.2
0.33
0.50
1.00
--
1.00
0.00
0.33
0.50
(2.5)
(5.5)
(12,5)

(12.5)
(1)
(2.5)
(5.5)
concentration with which
10.0
0.00
0.50
0.33
0.00
0.50
0,00
0.50
0.50
they
(2)
(7.5)
(4)
(2)
(7.5)
(2)
(7.5)
(7.5)

K were ranked.
                                    394

-------
                         TABLE F.4. RANK SUMS
Effluent
Concentration
(*)
0.32
1.00
3.2
10.0
Rank Sum


65
65.5
42
40
No. of
Replicates

8
8
7
8
Critical
Rank Sum

44
44
34
44
TABLE F.5. CRITICAL VALUES FOR WILCOXON'S RANK SUM TEST WITH
           aONFERRONI'S ADJUSTMENT OF ERROR RATE FOR COMPARISON
           OF "K" TREATMENTS VS A CONTROL FIVE PERCENT CRITICAL
           LEVEL (ONE-SIDED ALTERNATIVE:  TREATMENT CONTROL)
K No. Replicates No.
in Control

1 3
4
5
6
7
8
9
10
2 3
4
5
6
7
8
9
10

3
6
6
7
8
8
9
10
10
_ _
_-
6
7
7
8
8
9
of ReplicatesrEffluent Concentration

4
10
11
12
13
14
15
16
17
— —
10
11
12
13
14
14
15

5
16
17
19
20
21
23
24
26
15
16
17
18
20
21
22
23

6
23
24
26
28
29
31
33
35
22
23
24
26
27
29
31
32

7
30
32
34
36
39
41
43
45
29
31
33
34
36
38
40
42

8
39
41
44
46
49
51
54
56
38
40
42
44
46
49
51
53

9
49
51
54
57
60
63
66
69
47
49
52
55
57
60
62
65

10
59
62
66
69
72
72
79
82
58
60
63
66
69
72
75
78
                                 395

-------
TABLE F.5. CRITICAL VALUES FOR WILCOXON'S RANK SUM TEST WITH
           BONFERRONI'S ADJUSTMENT OF ERROR RATE FOR COMPARISON
           OF "K" TREATMENTS VS A CONTROL FIVE PERCENT CRITICAL
           LEVEL (ONE-SIDED ALTERNATIVE:  TREATMENT CONTROL)(CONTINUED)
m K No. Replice
mm in Control
mm.
H — _ 	
BESc o i
H ?
1
1
1
I z
H| o
m 10
II 10
H8|
1 4 2
i
i
1
i
I 9

B 10
msms
KBBy
1 5 2
H 4
1 i
i
1 0
I Q
1 •,»
B 10
B
Ii 6 J
i ; r-
1

i 8

i 9
i 10
i 	 	
tes No. of Replicates:Effluent Concentration

3 4
— —
10
11
6 11
7 12
7 13
7 13
8 14
- -
--
10
6 11
6 12
7 12
7 13
7 14
_ _ __
--
10
11
6 11
6 12
7 13
7 13
— - _„
--
10
11
6 Tl
6 12
6 12
7 13

5
_ _
16
17
18
19
20
21
22
— _
15
16
17
18
19
20
21
_ _
15
16
17
18
19
20
21
_ _
15
16
16
17
18
19
20

6
21
22
24
25
26
28
29
31
21
22
23
24
26
27
28
30
— —
22
23
24
25
27
28
29
— —
21
22
24
25
26
27
29

7
29
30
32
33
35
37
39
41
28
30
31
33
34
36
38
40
28
29
31
32
34
35
37
39
28
29
30
32
33
35
37
38

8
37
39
41
43
45
47
49
51
37
38
40
42
44
46
48
50
36
38
40
42
43
45
47
49
36
38
39
41
43
45
47
49

9
46
48
51
53
56
.58
61
63
46
48
50
52
55
57
60
62
46
48
50
52
54
56
59
61
45
47
49
51
54
56
58
60

10
- 57
59
62
65
68
70
73
76
56
59
61
64
67
69
72
75
56
58
61
63
66
68
71
74
56
58
60
63
65
68
70
73
i 	 	 	
i 396

-------
TABLE F.5. CRITICAL VALUES FOR WILCOXON RANK SUM TEST WITH
, uu... u>t-tu
-------
                                   APPENDIX  G

                                PROBIT ANALYSIS

 1.1   This program calculates  the  LC50,  LC15, LC10,  LC5,  and  LCI  values,  and
 associated 95% confidence intervals.

 2.   The program is written in IBM PC  Basic.for  the  IBM compatible  PC  by
 D.  L.  Weiner,  Computer Sciences Corporation, 26 W.  Martin  Luther King Drive,
 Cincinnati,  Ohio 45268.   A compiled version of  the  program can be  obtained
;from Computer  Sciences Corporation by sending a diskette with a  written
^request.

j2.1  Data  input is illustrated by  a set  of mortality data from a  sheepshead
.minnow embryo-larval  survival and teratogenicity test.  The  program begins
;with a request for the following  information:

     1.  Output  designation (P  = printer,  D = disk file).
     2.  Title for the  output.
     3.  A  selection of model fitting options {see sample output
        for a detailed description of  options).   If  Option  2  is
        selected, the  theoretical  lower  threshold needs to  be entered.
        If option 3 is selected, the program request the number of
        animals responding-in  the  control group  and  the total number
        of original animals in the control group be  entered.
:,    4.  The number of  test concentrations.

 2.2.   The program then requests information on  the  results at each
 concentration, beginning  with the lowest concentration.

     1.  Concentration.
     2.  Number  of organisms responding.
     3.  Total number of exposed organisms.

 2.2.1.  See  sample data input on  the  next page.
                                      398

-------
2.2.1 Sample Data Input.
        uuuuuuuuuuuuuuuuuuuuuuuuuuuw
        u                                                           u
        ij                 EPA PRCBIT ANALYSIS PROGRAM               U
        ij               USED FOR CAIXIJLATING DC VALUES              U
        U                         Version 1.4                       U
        UUUiJUUUUUUTOJUU^^
    Output to printer or disk file (P / D)? p
    Title ? Probit Analysis  of Sheepshead Minnow Rnbryo-Larval Data
            Model Fitting Options Which Are Available

    1) Fit a model which  includes two parameters: an  intercept and a
      slope.  This model assumes that  the spontaneous response
       (in controls)  is zero.  No control data are entered if this
      option  is specified.

    2) Fit a model which  includes three parameters: an intercept, a
      slope and a theoretical lower  threshold which  represents the
      level of spontaneous  response  (in controls).   This option
      requires the user  to  input the theoretical lower threshold
       (the value must be between 0.0 and 0.99).  No  control data is
      entered if this option is specified.

    3) Fit a model which  includes three parameters, an intercept, a
      slope and a lower  threshold.   The lover threshold is estimated
      based on control data which are  input by the user. If the number
      responding in  the  control group  is zero, then  this option is
      indentical to  option  two  (above).

      Your choice  (1, 2, or 3)? 3
   Number of responders in the control group = ? 2
   Number of animals exposed in the. concurrent control group = ?
   Number of administered concentrations ? 5
20
                                     399

-------
[2.2.1 Sample Data  Input  (Continued).
           Input data starting with the lowest concentration
          Concentration =  ? 0.5
          Number responding =  ?  2
          Number exposed = ? 20

          Concentration -  ? 1,0
          Number responding =  ?  1
          -Number exposed - ? 20

          Concentration =  ? 2.0
          Number responding =  ?  4
          Number exposed = ? 20

          Concentration =  ? 4.0
          Number responding =  ?  16
          Number exposed = ? 20

          Concentration =  ? 8.0
          Number responding -  ?  20
          Number exposed = ? 20

Number
1
2
3
4
5

Cone.
0.5000
1 . 0000
2.0000
4.0000
8.0000
Number
Resp.
2
1
4
16
20
, Number
Exposed
20
20
20
20
20
          Do you wish to modify your data ? n
          The number of control animals which responded
          The number of control animals exposed  =  20
          Do you wish to modify these values ? n
—  2
                                    400

-------
2.3  Sample Data Output

2.3.1  The program output includes the following:
        A table of the observed,  adjusted (using Abbott's formula)
        and predicted proportions responding at each concentration.
        Chi-square statistic for  heterogeneity.  This test is one
        indicator of how well  the data fit the model.
        Estimates of the mean  (mu) and standard deviation (sigma)
        of the underlying tolerance distribution.
        Estimates and standard errors of the intercept and slope of
        the fitted probit regression line.
        Estimate and standard  error of the lower threshold (if
        requested - requires control data on input).
        A table of estimated EC values and 95% confidence limits.
        Please note that EC, effective concentration, is  a broad term
        and applies to any response, such as fertilization,  death or
        immobilization.   If mortality data is entered in  the program
        as the response,  the EC estimates are equivalent  to  LC
        (lethal concentration)  estimates.
        A plot of the fitted probit regression line  with  observed
        data overlaid on the plot.
                                     401

-------
[2.3.2   Probit Statistics Output
  Probit Analysis of Sheepshead Minnow Embryo-Larval Data
      Cone.
Number
Exposed
ftfumber
Resp.
Observed
Proportion
Responding
Adjusted
Proportion
Responding
Predicted
Proportion
Responding
     Control
      0.5000
      1.0000
      2,0000
      4.0000
     8.0000
20
20
20
20
20
20
 2
 2
 1
 4
16
20
0.1000
0.1000
0.0500
0.2000
0.8000
1.0000
 Chi - Square Heterogeneity =     0.441
 Mu
 Sigma

 Parameter
0.479736
0.150766

Estimate    Std. Err.
 intercept
 Slope
1.818003    0.976915
6.632814    1.804695
 Spontaneous     0.084104    0.036007
 Response Rate
0.0000
0.0174
-.0372
0.1265
0.7816
1.0000
0.0841
0.0000
0.0007
0.1179
0.7914
0.9975
                                                95% Confidence Limits
     (    -0.096749,
     (     3.095611,

     (     0.013529,
                               3.732756)
                              10.170017}

                               0.154678)
       Estimated EC Values  and Confidence Limits
 Point
 EC 1.00
 EC 5.00
 EC10.00
 EC15.00
 SC50.00
 EC85.00
 EC90.00
 EC95.00
 EC99.00
   Cone.

    1.3459
    1.7051
    1.9343
    2.1061
    3.0181
    4.3250
    4.7093
    5.3423
    6.7680
           Lower       Upper
         95% Confidence Limits
          0.4533
          0.7439
          0.9654
          1.1484
          2.2676
          3.5656
          3.8443
          4.2566
          5.0712
                  1.9222
                  2.2689
                  2.4871
                  2.6523
                  3.6717
                  6.3827
                  7.5099
                  9.6486
                15.6871
                                    402

-------
J2.3.3   Plot of Adjusted Probits and Predicted Regression Line.
               PLOT OF ADJUSTED HABITS AND PREDICTED RH3*ESSICN LINE
        Probit
           10-*-
            2+
            0+0
             EC01
                           H	1	1	h
EC10    EC25     EC50     HT75     EC90
                                                                         EC99
                                           403

-------
[2.4  Listing of  Computer  Program for Probit  Analysis.
    10
    20
    30
    40

    50
    60
    70
    80
    90
    100
    1 10
    120
    130
    140
    150
    160
    1 70
    180
    190
    200
    210
    220
    230
    240
    250
    260
    270
    280
    290
    300
    310
    320
    330
    340
    350
    360
    370
    380
    390
    400
    410
    420
    430
    440
    450
    460
    470
    480
    490
    500
    510
    520
    530
    540
                                            output  is p r o b i t
                                            pv a Iues
                                         t er a t i on
   'EPA PROB1T  analysis  program :  Version 1.4  by ti.L.Weiner,  3/24/88
   'Written  for  IBM  PC  and  full  compatibles.   May requ'ire  minor
   'modifications  for  other  systems.
 DIM NDOSE(20),N8ESP(20),DOSE(20),LDOSE(20),CH1SO(18),T05(18)  OP<14>
 QY( 14 )                                                       '
 DIM UT(20>, YPROB(20),  WKY(20),  XPR]ME(20),  Y(20)
 DIM PK4) , 01 (4>,P2(8) ,02(8), P3(5),03<5),  PLTT$(51,71)
 B1GC=0!  :  PI  = 3.1415926535*
 GOSUB 720  ' input  dat a
 GOSUB 320  ' check  for  valid  data  file  type
  CLS : LOCATE  12,35  : PRINT  "Working  ...»              ...
  GOSUB 1970 '  read  in  t and  chisq  values
  1 sub 1950 is used  to  compute probits  :  input  is  no.r
 '  sub 2270 is  used  to  compute  area under  normal  curve
 FOR I =  1  TO  K
 W T < 1 ) = 1 !    '  initialize weights  to 0 or  1  fcr  first
 If NRESP(I) = 0  OR NRESP( I ) = NDOSE( I ) THEN  UT(I) = 0!
 IF BIGC  >= NRESP< 1 }/HDOSE< I }  THEN  WT(I)=0?
 XPR 1ME( !  )  = 0!
 NEXT I
 ITER = 0    '  begin  leroth  iteration
 GOSUB 3180 ' model  fitting
 GOSUB 3500 '  compute predicted probits  ,  working  profaits  (wky)
            1  and weights (wt>
 OK = 0
 GOSUB 3640 '
 IF OK=1  TKEM
 ITER = ITER'
 GOTO 210
REH  - subroutine to force  form feeds
PRINT #2,  CHR$(12)
RETURH
REM - check for valid data  files  ,  wt
N1=1 : N2=1  : N3=0  : WT( 1 ) =DOSE<1>
FOR I = 2  TO N
FOR J = 1  TO N2
                    THEN GOTO 390
 check  for  convergence
 GOTO  3950   '  convergence  achieveo,
•  1   '  begin  next  iteration
                                                  f
                         is  used  as  a  work  array  here
     I = 2
     J = 1
1 F DOSE( J )  =  WT(J)
NEXT J
N1 = N1 + 1 : UT(N1 ) = DOSE ( I )  :  N2 = ti1
NEXT I

FOR  1=1 TO  K
IF KRESP(I)>0 AND NRESPf I )   TKEN
WEXT I
IF N3<=1 THEN GOTO 620
N2 = N3
FOR  1 = 1 TO  N3 - 1
FOR  J=J+1 TO  N3
IF UT(1)=WT(J)  THEN N2=N2-1
NEXT J
NEXT I
N3 = N2
IF N1 >= 3  AND  N3 >=  2  THEN RETURN
IF N1 >= 3  THEN  GOTO  620
CLS  :PRINT  "  ":PRINT  "  "
                                      UT (N3)=DOSE( I )
                                          404

-------
2.4   Listing of  Computer  Program  for Probit Analysis  (Continued).
  550 PRINT "
  560 PRINT »
  570 PRINT "
  580 PRINT "
  590 PRINT »
  600 PRINT "
  610  IF N3>=2 THEN GOTO  5550
  620  CLS:  PRINT " ":PR1NT
  630  PRINT "
  640  PRINT "
  650  PRINT »
  660  PRINT "
  670  PRINT "
  680  PRINT "
  690  PRINT "
  700  PRIKT "
                 I I £ £ E E E E E E E E E I t E E E E E [ [E[ [ E E E E £ E IE £ E [ E £ E E E £ [£ £[ [ £ £ [ £ E E £ f E E E E E £'
                                                                               ['
                 [   Note  : data  file  must  contain at  least  three             [i
                 I          unique  concentration levels.                      ,,
                 C                                                             ['
                 t t U til t£E££ [££££££[ EEEEEEEE££[tEEE££[E£f [£[[[£££[ E [£££££££££'
                 II E f EE E E £ £ EE E EE EEEIE E££E EEf EEEEEE EEEEEEfE £ EUI ' I E £ EEEE [ £'EEEE'
                                                                               V
                 t    Note  : data  filemust  containat ieest  t we               ri
                 I           concentrations  for  K h i c ,h t^epercert              r'
                 E           responding  is  between  OX arc 10C%                 M
                 E 11 E E E f m E E E E i r n .m E M ; [ E E £ i E E E i; i r i E E : M : E :: E : E E E E E E E £ E m •
              : P R I N T  "  "
710 GOTO  5550
720 REH   •   Data  Input  subroutine
730 PR!(JT  "      I tEEEEEEEEIEEEEEEEEUtEE EMCEE £[[[[[ EEICEMd; ME E£E£EEEEEEE'
740PRINT"[
750 PRINT  "      [                   EPA  PROSIT  ANALYSIS  PROGRAM                 [.
760PRINT"      E                USED  FOR  CALCULATING  EC  VALUES                p
7 7 0 P R I N T  »      [                          V e r s * o n  1 . 4                          £,
780 PR!"T  "      (EEEEEEUEi £E I t £ I E E £ £ £ E £ t £ £ E I E £ £ [ [ £ E E E £ £ IE £ E M : I E t E I £ £ E E E £ E E £'
790 PRINT  "  " : PR I NT  "  "
800 INPUT  "Output  to  printer  or dis-k  file  (P  /  D)";ANS$
810 IF AWS$='""  TS*EV  G 07 0 8CD
820 ANS$ = LEFTt(AKSS( 1 >; I F ASC(ANS$}>96  THEN  ANS$ = ChRS{ASC"P"  AND  ANS$<>U'D"  GOTO  800
840 IF ANS$="P"  THEN  F 1 L E S ="Ip t1 :"
850 IF ANS$ = »D"  THEN  INPUT  "File  ns-ne  for  outpLt";  FILES
860 INPUT  "Title  ";TITLE$
870 OPEN  FILES  FOR  OUTPUT AS  #2
880 PRINT  #2,"  »
890 PRINT  #2,"                     EPA  PROSIT  ANALYSIS  PROGRAM"
900 PRIHT  #2,"                   USED  FOR CALCULATING  EC  VALUES"
910 PRINT  32,"                           Version  1.4"
920 PRINT  #2,"  ":  PRINT  #2,"  " : P R I N T  it 2, "  ":  CLS
                         Model  Fitting Options Which Are Available ni:PRIHT " "
                1)  Fit  a  model  which  includes  two parameters: a-i  intercept and
930 PRINT  '
940 PBINT  "
    a"
950 PRINT  »
960 PRINT  "
970 PRINT  »
980 PRINT  "
     a"
990 PRINT  "
     the"
1000 PRINT "
1010 PRINT »
1020 PRINT »
     is"
1030 PRINT "
1040 PRINT "
     a"
                    slope.  This model  assumes  tKat  the spontaneous response"
                    (in controls)  is  lero.   No  control  data are entered  if  this"
                    option is specified,";PRINT  "  "
                 2)  Fit a model which  includes  three parameters:  en intercept

                    slope and a theoretical  lower  threshold which represents

                     level of spontaneous  response  (in  controls).  This  option"
                     requires the  user  to  input  the  theoretical lower threshold"
                     (the value must  be between  0.0  and 0.99).   No control  data

                     entered if this  option  is  specified.":PRINT  " "
                  3)  Fit  a model which  includes  three  parameters, an intercept,
                                           405

-------
2.4   Listing  of Computer  Program for Probit Analysis (Continued).
 1 070

 1080
 1090
 1100
 1110
 1 120
 1 130
 1140

 1 150
 1 160
 1 170

 1 ISO

 1 190
 1200
 1210

 1220
 1230
 124C
 1250
 1260
 1270
 1280
 1290

 1300
 1310
 1320
 1330
 1340
 1350

 1360
 1370

 1380
 1390
 1400
1410
1420
1430

1440
1450
1460
1470
1480
                     slope and  a  lower  threshold.   The lower threshold is
                     responding  in  the  control  group is  zero,  then this opti
1050 PRINT "
     est imated"

1060 PRINT "        based  on ^ntrol data which are input by  the  user.  If  the
     number"
     PRINT "
     i s "
     PRINT "        indenttcal  to option two (above).":PRINT  " "
     INPUT »        Tour choice <1,  2,  or 3 ) » ; C H 0 S
     IF CHOI <> "I" AND CHCS  <>  •' 2 '• AND CHOJ <> » 3 » THEN CLS  : GOTO  930
     IF CMOS = "1"  THEN CLS :  GOTO  1240
     •IF C H Q $ = •' 3 "  THEN CLS :  GOTO  1160
     CLS : INPUT "Spontaneous  response rate ";BIGC
     IF B1G01I OR  BJGC<0.  THEN  PRINT  "Value must be  between 0 a-d 1 » : G 0 T 0

     GOTO 1240
     INPUT "Number  of  responders in the control  group = " - N R C T R i.
     IT NRC7RL<=0   THEN PRINT  "Number  must  be  greater than  0, please reenter"
     GOTO 1160
     INPUT "Number  of  animals  exposed  in the concurrent contrcl  group  - ...
     NCTRL                                                                 '
     IF  NCTRL  < 0 THEN PRINT » I n v a I i d  numbe r ,  please  r e e n t e r -• : G 0 T 0 1180
     IF  NRCTRL>NCTRL THEN PRINT  "The number  of  responders must be no greater"
     IF  NRCTRL>NCTRL THEN PRINT  "than  the  number  exposed,  please r e e n t  e r'- -
     GOTO  1160
     C  =  NRCTRL /  NCTRL   '  empirical control  resp
     B1GC  = C              '^itiaE  predicted control  resp
     1VPUT  "'Number  of  administered  concent rat icr-s "' • N
     * 1  =  N :  I f  NCTRL  >  0  THEN  N1  = Ni  +  f
     IF  N> = 3  THEN  GOTO  1290
     CLS
     PRINT  "Sot  enough  concert rat i on levels  to  fit  s  mode I"' : GO' C  55-50
     3F  N1>20  THEN  PRINT  '-Maximum number of  concentration   levels exceeded'"
     GOTO   5550
     CLS
     PRINT:PRINT»lnput  data starting with the  lowest  cone entration"•PR INT
     DOSE(O) =  0    ' dummy value
     FOR J  = 1  TO  N
     INPUT"Concentrations  " ; D 0 S E ( J )
     IF  DOSE(J)<=0  THEN PRINT  "Invalid concentration  -  please  ree-ter» • GOT 0
     1340
     IF  DOSE(J> > DOSE(J-1)  THEN  GOTO 1390
     PRINT  :PRINT "Concentrstions must  be entered in ascending o^der  -  tow to
    h i g h . »
    PRINT  "Please  reenter the  data.":GOTO 1310    '-
     LDOSE(J) = LOG(OOSE(J ) )/2.3025851*  MoglO dose
     INPUT  "Number  responding  =  ";NRESP(J>
     IF  NRESP(J)<0  THEN PRINT  "Invalid number - please  r e en t er •': GO T 0  1400
     INPUT  "Number  exposed = ";NDOSE(J)
     IF  NRESP(J)>NOOSE(J)  THEN  PR I NT"Number responding can't  be greater than
    number exposed  • please reenter":  GOTO 1340
    PRINT
    NEXT J
    CLS
    PRINT "                           Number     Number
    PRINT "    Number        Cone.    Resp.      Exposed   "
                                           406

-------
 2.4
Listing of  Computer  Program for Probit Analysis  (Continued).
  1 490
  1500
  1510

  1520
  1530
  1540
  1550
  1560
  1570
 i1580
 ft 590
 ;1600
  1610

  1620
  1630
  1640
  1650
 1660

 1670
 1680
 1690
 1700
 1710
 1720
 1730
 1740
 1750
 1760
 1770

 1780
 1790
 1800
 1810
 1820
 1830
 1840
 1850
 1S60
 •1870
 |
[1830

[1890
j1900
-1910

,1920
|1930
 1940
i1950
                                        ntfan
                                                   ####";I;DOSE{I),
                                                                    : GOTO
 PRINT "
 FOR 1 = 1 70  N
 PRINT USING"       # #
 NRESP( I ),NDOSE( I )
 NEXT 1
 PRINT: INPUT  "Do you  wish  to modify your data " ; A N S S
 IF  A N S * = " " THEN  GOTO  1530
 ANS$ = LEFT$(ANSS, 1 } : I F  ASC(ANS$)>96 THEN ANS$ = CHRJ(ASC(ANS$) - 32>
 IF  ANS$<>"N" AND ANS$<>"T"  THEN GOTO 1530
 IF  ANS$="N" THEN GOTO  1680
 INPUT  "Observation  Number  to be modified "; 10 B
 IF  I08< = 0 OR  IOB>N  THEN  PRINT"  Invalid Kumber":GOT 0 1535
 INPUT  "Concentration  =  ";DOSE{IOB>
 IF  OOSE(10B}<=0 THEN  PRINT  "Invalid concentration -  p esse  -eenter
 1600
 LDOSE( IOB) = LOG(COSE( I 0B})/2.302585 1 #  'IcglC dcse
 INPUT  "Number responding =  ";KRESP96  T H E H A S S 1 = C ri R S C A S C ( A *, S S 3 - 3 2 >
 IF  ANSS<>"N"  AND ANS4<>"'Y" THEN GOTO 1710
 IF  ANS$="N»  THEN GOTO 1950
 INPUT  "Spontsneous  response  rate ";B1GC
 IF  BIG01!  OR B1GC<0!  THEN PRINT "Value must  be  between  0  and  l "-GOTO
 1760
 GOTO  1950
 PRINT  "The  number of control animals which  responded  =  ";kRCTRL
 PRINT  "The  number of control animals exposed   =  " ; KCTRL
 INPUT  "Do you  wish  to modify these  values  ";ANS$
 IF ANSt="»  THEN GOTO 1680
 ANS$ = LEFT$(ANS$, 1): If  ASC(ANS$)>96  THEN  ANS$ = CHR$tASC(ANSS3 - 32)
 IF ANSSo»N"  AND  ANSS<>"Y" THEN GOTO 1680
 IF ANS$="N"  THEN  GOTO 1950
 INPUT  "Number  of  responders  in  the  control  group  = ",-NRCTRL
 IF NRCTRL< = 0   THEN  PRIVT "Number must  be  greater  th3r.  0, piease reenter"
 GOTO 1860
 INPUT  "Number  of  animals exposed in  the  concurrent control  group =
 NCTRL
 IF NCTRL <  0  THEN PRINT "Invalid number,  please reenter":GOTO  1880
 IF NRCTRL>NCTRL  THEN PRINT  "The number  of  responders must  be no greater"
 IF NRCTRL>NCTRL  THEN PRINT  "than the number exposed, please  reenter":
GOTO 1860
C = NRCTRL /  NCTRL   '  empirical control  resp
BIGC 3 C             '  initial predicted  control resp
GOTO 1680
CLS
RETURN
                                           407

-------
,4  Listing of  Computer  Program for Probit Analysis  (Continued).
  1970
  1980
  1990
  2000
  2010
  2020
  2030
  2040
  2050
  2060
  2070
  2080
  2090
  2100
  2110
  2120
  2130
  2140
  2150
  2160
  2170
  2180
  2190
  2200
  2210
  2220
  2230
 2240
 2250
 226C
 227G
 2280
 2290
 2300
 2310
 2320
 2330
 2340
 2350
 2360
 2370
 2380
 2390
 2400
 2410
 2420
 2430
 2440
 2450
 2460
 2470
 2480
 2490
 2500
 2510
     subroutine  to toad in op, o y,  t 0 5 ,  chisq
 FOR I =  1  TO  9
 READ OP( 1 )  '  op
 NEXT 1
 DATA .01 , . 05, . 1
                 values  are  percent i I e s  for predicted curve

                 .15, .5, .85, .9, .95, .99
                 oy  values are probits corresponding  to  Op  values

                                      6.0364
                                      ( a I pha = O.C5 t wo- s
                                                  c f freedom
                                                              ( df )
 FOR  I  =  1  TO
 READ  OY( I )
 NEXT  I
 DATA  2.6737,3.3551,3.7184,3.9636,5.
 DATA  6,2816,6.6449,7.3263
  '   TOS  values  are 97.5 pereentiles
     for  a  t-distributio.n. I  denotes  the  degrees
  1   and  runs  from Idf to 18df
 FOR  I  =  1  TO  18
 READ  T 05 (.1 }
 NEXT  I
DATA  12.706,4. 303,3. 182,2, 776,2.571,2.447,2.365, 2. 3C6
DATA  2.262,2.228,2.201 ,2. 179,2.16,2.145,2. 131 ,2! 12,2. 11,2. 101
    CHISQ values  are  95  pereentiles  (alpha  =C,05  two  sided)
  1  for a chi-squa'e  distribution. I denotes  the  degrees  of  freedom
  '  (df) and  runs  from  Idf to ISdf.
FOR I = 1 70  18
READ CM I S0< I )
NEXT  I
DATA 3.841,S.991,7.815.9.488,11.07,12.592,14. fl'67, 15.5:7
DATA 16.919, 18.307,19.675.21.026,22.342.23.685,24.9
DATA 27.587,28. 869
RETURN
'  subroutine  to  co-put  p r o b •" t s
'  input is nd,r  * h e r e  nd = *exposed and
1  output = pr ob i t
                                                     •6 , 2* . 296
   precision equals that of  BASIC
P  =  R/ND
PPP=PP
If R=0   AND  1TER=0
IF R = ND  AND  I TER = 0
IF BIGC  >=  P  THEN
           PP=  (P  -  BIGC >/( 1 !
                                 BIGC)
                                         = #respor.din3
                                          Abbotts
                     THEN  XP = - 105 :  GOTO 2570
                     THEN  XP= 105 :  GOTO 2570
                  XP=- 1 05  :  GOTO 2570
                                       •nfinfty  if
                                       nf i rit y  'f
P0=-.322232431088*
P1 = - 1 !  :  P2=-.342242088547*
P4=-.0000453642210148*
Q0=.099348462606* :  Q1=.588581570495*
02  =  .531103462366*  :  03=.10353775285*
04=  .0038560700634*
XP = 0
IF  PPP > .5  TKE»  PP = 1 • PP
IF  PP=.5  THEN  GOTO 2570
Y =  SQR(IOG<1/(PP*PP»)
                               P3=- .02C423121G24S*
M
T2
T3
T4
U1
     (T*P4
     (T1  +
      *
PI ) «
PO )
+ Q3)
                                         408

-------
12.4  Listing of  Computer Program for  Probit  Analysis  (Continued).
          2520
          2530
          2540
          2550
          2560
          2570
          2580
          2590
          2600
          2610
          2620
          2630
          2640
         2650
         2660
         2670
         2680
         2690
         2700
         2710
         2720
         2730
         2740
         2750
         2760
         2770
        . 2780
         279C
         2SOO
         2810
         2823
         2830
         2840
         2850
         2860
         2870
         2880
         2890
         2900
         2910
         2920
        2930
        2940
        2950
        2960
        2970
        2980
        2990
        3000
        3010
        3020
        3030
        3040
        3050
        3060
  U2
  U3
  U4
  XP
  1 F
= ( U1
= (U2
= (U3
            U4 )
            XP=-XP
                           integral
                           pvaIue
                               of  normal  distribution
                        PU2)»21 .97926161894152*
                        PU4) = -3.56098437CT815390
                        Q1(2)=91 . 1649054C45T49*
             02)
             01)
             00 )
     =  r  +  = 0 THEN 2=  Z/SOR(2)
 P1<  1 ) = 242.6679552305318* :
 P1(3)=6.996383488619135# :
 OK  1 } = 215.0588758698612* :
 01 (3) = 15. 08279763040779* :
 P2< 1 ) = 300.4592610201616* ;
 P2<3)=339.3208167343437* ;
 P2<5 )=43. 16222722205673* :
 P2(7)=.564195517478974*  :
 G2<1) = 300.4592609569833*
 Q2(3)=931.354094S506096#
 02(5}=277.5854447439876*
 Q2{7)=12.78272731962942*
 P3(1)=-2.996107077035422D
 P3<3) = - .2269565935396869*  :
 P3(5) = -2. 23192459734 1.847S-02
 03C}=1.062C92305284679D-C2  :  C3(2)».19T3C89261G782SB*
 fl3(3)=1.051675107C67932#  : Q3<4>=1.98733201817T3<3*
 Q3(5 } = 1#
 CONST*.3939422804014327*
 1 F 2  > 4 #  G 0 T 0 3 0 1 0                              ...
 IF Z  >=  .46875* GOTO 2920
 TOP=P1(1):BOT=01(1)
 FOR  1  =  2  TO  4                                    '.'•'•".
 1 1  = 2 *  ( I - 1 )  :  z 2 = Z '  I 1                         •; :';
TOP  =  TOP  +  P1(I) « 22
BOT=BOT+01(I)*Z2
NEXT!                                               • •' •  -
R 1  =  TOP / BOT  :  ERFX =-2 ' fi 1
TOP =  P2( 1  )  : BOT = Q2( 1 )
FOR  I = 2  TO 8
           22  = 2  *  I 1
            P2( I ) *  ZZ
            02(I) *  22
                                                        02
                             P2(2)=451.9189537118729*
                             P2{4)=152.9892850469404*
                             P2(6)=7.21T758250883094*
                            P2(8) = -1.3686485 73 S27U7D-07
                             02(2 ) = 790.950925327898*
                             02(4)=638.9802644656312*
                             C2(6)=77.00015293522947*
                             02(8)=1#
                            03 : P3(2)=-4.947309t06232507D
                              P3(4}=-.27S6613CS6C9647S*
                                                       02
                                   GOTO  3MO
 It  =  1-1
 TOP =  TOP
 BOT =  BOT
 NEXT  !
 R2  =  TOP /  BOT  :  EHFCX = 0*
 IF  2  <  13.038*  THEN ERFCX = EXP(
 ERFX  =  1*  -  ERFCX  : GOTO 3110
 22  =  1* /  (2*2)  :  TOP = P3(1> :  BOT
 FOR I  = 2  TO  5
                             2*2)
                                         R2
                                = 03( 1 )
11 = 2 *
TOP=TOP
NEXT 1
R3 = TOP
    (I - 1 )  :  22
    • P3
-------
1.4   Listing of Computer Program for Probit Analysis (Continued).
1
1
I
1
i
1
1
i 3070
1 3080
i 3090
1 3100
1 3110
• 3120
iEm 3i3°
mm 3i4G
H 315°
Hi 316°
asm 3170
HR 3180
I^BHB
Hg| 3190
|BJE 3200
ffiS 3210
II 3220
ra| 3230
IH 3240
SlU 3250
iBBB
OH 326°
IH 32?°
EH 3230
IH 329°
IB 330°
HH 3310
IM 3320
I
»ffl
i
P
i
i




1





1
1
(81
1
I

I
L
3330
3340
3350
336C
3370
3380
3390
3400
3410
3420
3430
3440
3450
3460
3470
3480
3490
3500
3510
3520
3530
3540
3550
• 3560





I 3570
I 3580
1 359°
1 3600
W
i 3610
CONST = COHST * SQR<2#) : El = COHST * X2 • R3
ERFCX = Q#
IF 2 < 13.038* THEN ERFCX = ( E X P ( - Z * Z > / 2 ) * E1
E R f X = 1 # - E R F C X
E R F D U = E R F X - ,
IFZ1=OTHENPVALUE=(1+ERFDy)/2!
Z=Z1'recoverz
IF RVALUE <=0 THEN RVALUE =. 000001
IF RVALUE >=1 THEN RVALUE = .999999 : '
RETURN ,
1 subroutine to fit probit model
SNW = Ot : SNUX=0! : SNUY=0! : SNWXY=C! : SNUXX=Q! : SNWYY=G!
SNUXP = 0 > : SKWXPP = 0! : SNUXPY = 0! : SNWXXP = 0!
FOR I s 1 TO M
ND = NDOSE(I) : W = WT{!) : X = LDCSEC1) ; fi = NRESP(I)
NW = ND*W : XPR = XPRIME(l)
IF 1TER <=0 THEN GOSUB 2270 : YPROB(!}=PROBIT : Y=TPKOB 0 THEM Y = WKY(I) ' working probit if not 0 iteration
SNUY = SNWY + NU»Y : SHUYY = SNWYY + KU*Y*Y : SNWXY = SWWXY + N U * X - Y
SVW = SNW '+ NW : SNWX = SKUX + KW*X : SMUXX = SNWXX + KW»X*X
SNUXP = SNWXP + NU-XPR : SMWXPP = SNUXPP + MW*XPR*XPR
SNWXPY = SNWXPY + MW'XPR*Y : SVWXXP = SNUXXP + KW * X * XPft
NEXT 1
X8AR = SNWX/SNH ; YBAR = SNUY/SNU ; XPBAR = SHWXP / SNW
SYY = SNWYY • SNWY * SNWY / SKW : SXX = SNWXX • SKUX * SNWX / SNU
SX.Y = SNWXY - iKWX * SNWY / SNU : SXPXP = ShWXPP - SNJXP * SNWXP / SNU
SXXP = SNWXXP • SWWX * SNUXP / SNU
SXPY=SMWXPY-SNUXP»S«UY/SNU
I f SXX = 0 T KEN GOTO 5610
SLOPE = SXY /SXX
1STCPT = YBAR - SLOPE * XBAR
IF NCTRL <= 0 THEN RETURN ' otherwise adjust for natural response rate
SXPXP = SXPXP + NCTRL * (1! - BIGO/BIGC
SXPY = SXPY + NCTRL * (C • BIGO/BIGC
NUK = (SXY/SXX) • ((SXXP " SXPY) / (SXPXP * SXX})
DEN - 1' • ((SXXP * SXXP) / (SXPXP * SXX))
SLOPE = NUH / DEN
TEMP = SXY / SXX ;
DELC = (1! • BIGC) * ((SXPY - TEMP * SXXP) / (SXPXP • ( S X X => * S X X P ) / S X X ) )
BIGC = BIGC + DELC
INTCPT = YBAR - SLOPE * XBAR - (DELC * XPBAR / (If - BIGC))
RETURN
1 subroutine to compute y, wky, w, zv : input is n, tdose, intcpt, stope
FOR K = 1 TO N
Y(K) = INTCPT + LDOSE(K) * SLOPE ' predicted probit
2 = Y(K) - 5) : ' § 0 S U B 2590 ' get p value ,
2V = <1!/(SOR(2)*PI)))*EXP(-. 5*2*2)
I F 2V< = 0 THEN 2V= .000001
P = NRESP(K)/NDOSE
-------
2.4   Listing  of Computer Program for Probit Analysis (Continued).
 3620
 3630
 3640
 3650
 3660
 3670
 3680
 3690
 3700
 3710
 3720
 3730
 3740
 3750
 376C
 3770

 3780

 3790

 3800

 3810

 3820

 3830"

 3840
 3850
 3860
 3870
 3880
 3890
 3900
 3910
 3920
 3930
 3940
 3950
 3960
 3970

 3980

 3990

 4000
 4010
4020
4030

4040
4050
 NEXT  K
 RE TURN
 1  subroutine to check for- convergence
 IT MER  >  25 THEN GOTO 3750   - modify here if more  iteratlons  required
 IF  1TER>0  THEN GOTO 3690
 OLDINT  =  1NTCPT ; OLDSLO = SLOPE
 GOTO  3940
 IF  {OLD1NT=0>  OR 
-------
 2.4   Listing of Computer  Program  for Probit  Analysis  (Continued).

  4060  RATIO  = NRESP(L)  /  NDOSEU)  :ARATIO  = (RAT 10 -B1GCw, i ,  .1
  4070  PRIHT  #2. USING  «*«**#.*#„    „,„    ( R" 1 0 B , 6C ,/  1 , . B , cc j.
                                     #S##

  4080  NEXT  L
  4090  CHIHET  =  STY  -  SLOPE * SXY
  4100  IF  NCTRL>0  THEN  CHIHET = CHIHET  -  DELC
  4110  PRINT  #2,  '•  »:  PRINT #2,  	PRINT
  4120  PRINT  #2,  USJNG  "###.###";CHIHET - PRJ
  4130  NDF =  N  -  2  :  T •=  1.96 :  HET = 1 ! '
  4140  IF  CHIHET  <=  CHISQ  THEN GOTO  4240
  4150  T = T05 C ,,E, SES  B   ,,  J  (SX)(
 4^6_ G= HET » T*T * SEB / C S L 0 P £ * S L CP E )
 4270 If G < 1 ' THEN GOTO 4350
 4280 PRINT 82 , ""*"-"'»''***»•»****»«»*
4290 PRINT H2,  "*
     * 11

4300 PRINT #2,  "*

4310 PRINT »2,  "«
     * it

4320 PRINT *2,  »*
     * n

4330 PRINT it 2,  ''*<
                       Slope not ifgBlMe.nt|y -Hffr.Bt


                       EC fiducial  limits cannot be computed.
4340  PRINT  #2,  »  »
4350  IF  NCTRL  > 0  THEN  SEC  =  (1!  -  BIGC)*2   /  (SXPXP
4360  SEI  =  (11/SNW)  +  (X8AR *  XBAR  *  SEB)
4370  IF  NCTRL  > 0  THEN   SEI =  SEI  *  
-------
j2.4  Listing of Computer Program for Probit Analysis  (Continued).
   4440  PRINT  #2,  USING "Intercept    ####.#*####
   4450
   4460

   4470
   4480
   4490
   4500

   4510
   4520
   4530
   4540
   4550
   4560
   4570
   4580
   4590
   4600
   4610
   4620
   4630
   4640
   4650
   4660
   4670
   46SO
   4690
   4700
   4710
   4720
   4730
   4740
   4750
   4760
   4770
   4780
   4790
   4600
   4810
   4815
   4820
   4830
   4840
   4850
   4860
   4870
   4880
   4890
   4900
   4910
   4920
                                               SCR{SEC*H£ T >
XL = SLOPE -  T * SQRCSEB»HET) :  XU = SLOPE + T  *  SGR(SEB-HET)
PRINT #2, USING "Slope        ####.###### #«#*.##*###     <#####.######
 *«###.######)»;SLOPE,SOHCSEB*KET),XL,XU
F R I N T * 2 , " "
IF NCTRL <=0  THEN GOTO 4530
XL = B1GC •  T * SQH(SEC*HET) :  XU =  B1GC
PRINT #2, USING "Spontaneous  *###.#*##»«
 #####.#*####)";BIGC,SOR(SEC«HET),XL,XU
PRINT #2, "Response Rate"
GOTO 4550
PRINT #2,
PRINT #2,
GOSUB 290
PRINT #2,
PRINT #2,'
PRINT
          "Theoretical Spontaneous Response Rate  =
          USING "#,####"; B I C C
      #2 . «
PRINT  #2, "Point
PRINT  * 2 ,  "'  "
                PRINT #2, TITLES ;  PRINT  32,  "  «
                Estimated EC Values and Confide nee  L ' SIM s
                                                L owe r
                             Cone .
                                              95%  Confidence
                                                             PRINT #2,
                                                            Upper"
                                                             Limits"
FOR I
H = (
IF G> =
       =  1  TO  '
       OY{ I )  -
       1 !  THEN
 IF  NCTRL  >
 T EKP  = H  +
 SE  =  SGR  (
 XL  =  TEMP
 XU  =  TEMP
           0!
           (G
           < 1
            T
          •  T
1NTCPT }  / SLOPE
GOTO 4900
THEN GOT 0 4730
/  < 1 !  -  G»*CH •
 •  G >/SKW * £(K-
*  SE / (SLOPE  *
*  SE / (SLOPE  *
 XBAR )
XBAR }*2 )*SE8
'M »  '  G ) }
(1 »  -  GM
                                               SE  =  SE
                                                         SCR ( HET )
 IF XL<-10  THEM XL = • 1 0
 IF XU> 20  THEN XU=  20
 GOTO 4920
 1  fixup  formulas  if  a  threshold  (spontaneous response) was estimated
 C11 = SEB
 TEHPC =  «1!  - BIGC) "  (1!  - BIGC»
 C22 = SE C  / T EKPC
 C12 = 11 /  (SXXP  -  (SXX  * SXPXP)/SXXP>
 R1 = H +  20  THEN XLM  20  :
 IF G<1I   THEN  GOTO 4920
 PRINT 02,USING "EC##.##
 GOTO 4930
 PRINT #2,USING "EC##.##
                                                       10-H
                                    ########.####
        ##########.####>
  4930 NEXT I
                   100*0P(I ) ;  10'M;  10*XL;  10"XU
                                          413

-------
(2.4
      Listing of Computer Program  for Probit Analysis (Continued).
  4940 IF HSG=1  THEN PRINT #2,  -' --:  PRINT  #2,»  NOTE
        or equal  to 1.E20 are really infinite"
       GOSUB  290
       'subroutine  to  do the probit  plot
  4950
  4960
  4970
  4980
 4990
 5000
 5010
 5020
 5030
 5040
 5050
 5060
 5C70
 5080
. 5090
 51CO
 5110
 5120
 5130
 5140
 5150
 5 160
 5170
 5132
 5 190
 52CO
 5210
 5220
 5230
 5240
 5250
 5260
 5270
 5280
 5290

 5300
 5310
5320
5330
5340
5350
5360
5370
5380
       PRINT
       PRINT
        L IKE
       PRINT
        # 2,  TITLES  :
        #2,  »
         :  PRINT  #2,
        ft 2 ,  "Probit"
                    PRINT # 2,  » «
                    PLOT OF ADJUSTED PROBITS  AND  PREDICTED REGRESSIONS
                                       LLD99 =
  D R 0 W = 1 0 >  :  RADJ  =  .2
  LLD01  =  (2.6732-INTCPT)/SLOPE
  ADJ =  CLLD99  •  LLD01 )/  68
  DCOL  = LLDQ1  :  CADJ =  ADJ
  FOR J=1  TO  H
  RDIFF  =  DROU  -  YPROB(J)
  i RQWX  =  RDIFF/RADJ  +  1
  CD I F F  =  LOOSE( J ) •   DCOL
  1 COLX  =  CD I  F F/ CADJ  -*•  1
  IF YPROB(J)  < 0 THEN  IROWX=51
  IF YPROB(J)  > 10 THEN  IROUX=1
  IF LOOSE(J)LLD99 THEN  ICOLX=71
  PLTT$<]R0yX,ICOLX)=»o»     •  change plotting
  KEXT  J
  NO= 1 00   ' number cf points on
  FOR 1=1 TO 99
  R= I
               . C  T fi E H G-0 T 0 5 3 C 0
               c e » p u t e  p r c fc ; t
                i NT CF ! } /SL CP E
                PROB r T
                      1
                                              <7.327- IN7CPT}/S,OPE
                                                               if desired
                                         predicted  curve
                                        i e
                                             P  is  negative
                                      EC  value
 IF  I / », D  < =  £
 G3S.JB 2270
 L CGD=( PRGB1T
 RDIFF =  DRCW
 IROUX =  RD! F F/RADJ
 CD I F F =  L 0GB •  DCOL
 ICOLX =  CD I FF/CADJ + 1
 IF  PROBIT  < C THEN IROUX=51
 IF  PROBIT  > 10  THEN !ROWX=1
 IF  LOGO  <  LLD01  THEN ICOLX=1
 IF  LOGO  > LLD99  THEM JCOL%=71
 IF PLTTt(IROWX,ICOLX)  <>»o» THEN PLTT$(1ROWX,IC0LX}-
 plotting symbols here if desired
NEXT I

FOR  1=1
                                                                change
                             ( 5 * J J X )
                             USING »
5390
5400
5410
5420
5430
5440
        TO 51
 11=1-1
 J J X = ( I I - 2 ) / 5 : J J X = I I -
 3 F JJS = 0 THEN PRINT #2,
 IF JJXoO THEN PRINT  #?, »
 FOR J = 1  TO 71
 IF < P t T T $ < 1 , J ) <> " o •• AND P L T T $ < I , J )  < >
 plotting symbols here if desired
 JJ=J-1-LAG :  IF JJ <= 0  THEN GOTO  5430
 FOR K=1  TO JJ
PRINT #2,  " ";
NEXT K
PRINT #2,  PLTTS{I,J);
LAG = J
                                                PRINT  #2,"*";;lfUH =
                                             "."  )  THEN GOTO 5450  '  change

                                             'print blanks out to next symbol
                                        414

-------
G.4  Listing of  Computer  Program for Profait  Analysis  (Continued).
   5450  NEXT  J
   5460  PRINT  #2,  "  "
   5470  LAG=0
   5480  NEXT  I
   5490  BOT$="•+-----
 5500
 5510

 5520
 5530
 5540
 5550
 5560
 5570
 558G
 5590
 5600
 5610
 5620
 5630
 5640
 5650
 5660
 5670
 5675
 5680
 569D
 57CO
 5710
 5720
 5730
 5740
 5750
 5760
 5770
5780
5790
5800
5810
5820
5830
        PRINT  tf 2 , "
        BOT$=»EC01
                        " ; B 0 T $
                            EC 10
                   EC 99"
                        ";BOT$
                                      EC25
                                                EC50
                                                          EC75
                                                                   EC90
PRINT # 2,"
GCSUB 290
IF FlLES-o-Mptl :"  THEN  LOCATE  12,30  :  PRINT  "Output  stored  in  ",-FILES
LOCATE 15,1  ;IN.PUT  "Fit  another  data  set  ";AKS$
IF ANSS-"" THEN GOTO 5550
AVSS = LEfT$(ANSS, 1 ) : 1 F ASCCANS$)>96 THEN AKS$ = CHR$(ASC"S" AND  ANS$«>"Y"  THEN  GOTO  5550
IF ANS$="Y"  THEN  CLEAR  : CLOSE #2  :  CIS : GOTO 4C
GOTO 5910
CL S
P R ! K T " " ; P R I H T »  »
PRINT"      [
PRINT"      [
PRINT"      [   Mote
PRINT"      {
PRINT"      E
PRINT"      E
P R I K T "      [
                           EEC [ [[ [[ [[[ECEUEEEE [mm [[[[ E [[I E mmt [I
                               iterations are not  converging.   This usually
                               means that only one concentration  is on the
                               linear  portion of  the  concentration response
                               curve.
       f S I N T
                      E :: f m E m E E E m E i E m m m E i; m H m m :
                                                                        1 1 1 tu t
        IF WCTRL  =  G  ThEN  GOTO  5S10
        P.RINT:PRINT  -it  may  be  possible  to  fit  the  data  assuiti^g  the  spontaneous"
        PRINT  "control  rate  is  zero.":PRlKT
        INPUT  "Would  you  like  to  try";ANS$
        IF ANSS=""  THEN  GOTO  5740
        ANS$ = LEFT$(ANS$, 1): J F  ASC(ANS$»96  THEN  A N S $ = C K R $ ( A S C ( A K S J ) - 3 2 )
        IF ANS$o»N"  AND  AKSSo"Y"  THEN  GOTO  5740
        IF ANS$="N" THEN  GOTO  5810
        NCTRL=0:NRCTRL=0:BIGC=0
        CLS :  LOCATE  12,35  :PR!NT  "Working  ...";  GOTO  140
        PRINT  #2, TITLES: PRINT  *2,  ""
        PRINT  #2,"  ":PftINT #2," "
        PRINT  #2,"    **»"**"»««•««****»»»***«**»**«*»**...»«,*,,,..***,.,*»*.,.,
; 5840  PRINT  #2,"
       * ii
 5850  PRINT  #2,"
       * ii
15860  PRINT  #2,"

 5870  PRINT  #2,"
       * N
 5875  PRINT  #2,"

 5880  PRINT  #2,"
       * it
                        Note  :  Iterations ere not converging.  This  usually

                                means  that only one concentration is  on  the

                                linear portion of the concentrstfon response
                                          415

-------
2.4  Listing of Computer Program for Probit Analysis (Continued)
5890  PRINT fl2,

5900  PRINT
5910  END
              : P R I N T " " :  GOTO 5550
                                     416

-------