9637

• UCREFRP Catalog Number: 9637
• Author: American Society For Testing And Materials.
• Title: Standard Practice For Conducting Acute Toxicity Tests With Fishes, Microinvertebrates, And Amphibians.
• USFW Year: 1980.
• USFW - Doc Type: Philadelphia, PA.
• Copyright Material: YES

lutions should be checked against other stan-
dard sol~tions whenever possible. An analytical
method IS unacceptable if likely degradation
pr~du~ts of the toxicant, such as hydrolysis and
?xldatlOn products, give positive or negative
mterferences, unless such degradation products
are shown to be absent from the test chambers
during the test. In general, atomic absorption
spectrophotom~tric methods for metals and gas
chromatographIc methods for organic com-
pounds are preferable to colorimetric methods.
In addition to measuring the total concentra-
tion of toxicant in the test chambers, measure-
ment of either the "dissolved" or the "undis-
solved" fraction of the toxicant is desirable.
Especially for inorganic substances the "dis-
so~ved" fraction is usually defmed and deter-
mmed as that which passes through a 0.45-f.lm
membrane filter. Glass filter holders are best
for organic toxicants, but plastic holders are
best for metals. Filters and their holders must
always be prewashed by filtering distilled water
through them and then filtering a portion of
the solution of interest before the fmal fJltration
is performed. The final portion of the filtered
distilled water should be analyzed to ensure
that the filter is not contaminated with toxicant.
The sample must be filtered within 30 min after
it is taken from the test chamber. Whenever
samples from a toxicity test are analyzed, at
least one reagent blank must also be analyzed
if appropriate. Also, at least one sample for th~
method of known additions must be prepared
by adding toxicant to water from a control test
chamber to match the next to the lowest toxi-
cant concentration used in the toxicity test.

12. Quality Criteria

12.1 A test is unacceptable if more than a
total o~ 10 % of the control organisms appear
to be dIseased or stressed or die in a test deter-
mining an LC50 or appear to be diseased or
stressed or die or show the effect in a test
determining an EC50.
12.2 A test must meet both of the following
criteria so that the LC50 or EC50 can be cal-
culated with reasonable accuracy:
12.2.1 The concentration of toxicant in each
treatment, except for the highest concentration
and the control treatment(s), must be at least
60 % of that in the next higher one.
12.2.2 One treatment other than the control
should have killed or affected less than 37 % of

~ffi~

E 729

the organisms exposed to it, and one treatment
should have killed or affected more than 63 %
of the organisms.

13. Calculations

13.1 For each set of data the LC50 or EC50
and the 95 % confidence limits must be calcu-
lated on the basis of (a) the measured initial
concentrations of toxicant, if available, or the
calculated initial concentrations for static tests
and (b) the average measured concentration~
of toxicant, if available, or the calculated av-
erage concentrations for flow-through tests. If
other LC or EC values are calculated, their 95
% confidence limits must also be calculated.
13.2 Most acute toxicity tests produce quan-
tal data, that is, counts of the number of items
in two mutually exclusive categories, such as
alive or dead. A variety of methods (26) can be
used to calculate an LC50 or EC50 and 95 %
confidence limits from quantal data that con-
tain two or more concentrations at which the
percent dead or affected is between zero and
100, but the most widely used are the probit,
moving average, and Litchfield-Wilcoxon. The
binomial test can usually be used to obtain
statistically sound information about the LC50
or EC50, even when less than two concentra-
tions kill or affect between zero and 100 %. The
binomial test does not provide a point estimate
of the LC50 or EC50, but it does provide a
range within which the LC50 or EC50 should
lie. To ensure that calculations are performed
correctly, the hypothetical sets of data in Table
8 should be analyzed. The results should fall
within the ranges of acceptable values given
~herem. These ranges were derived by compar-
m~ the valu.es calculated by various people
usmg a vanety of methods of calculation.
Ranges are given because there are no single
correct values. When valid methods of calcu-
lation are correctly used, values within the
ranges given should be obtained.
13.3 Although they generally require more
effort to obtain, quantitative data on individual
organisms, such as time-to-death or shell
g~owth, contain more information per orga-
msm than do quantal data. Quantitative data
are usually analyzed by a regression technique
after use of a transformation if appropriate
(27).

14. Reports

14.1 The record of the test and published

16