Basic HTML Version
MERC STDN Test Plan
10
G9 document in Section 5.2.4 (IMO, 2008). The algae test represents a true population growth
test.
In addition to the ballast water discharge efficacy testing, sampling and toxicity testing of
the water in the test tank will also be conducted on a minimum of one of the six test trials.
Sampling will be done at the time of tank filling/treatment, and at one day and five days. The
analytical and bioassay methods described above will be used to analyze these different time
point samples.
Statistical Analyses:
Toxicity endpoints will include survival in acute fish and invertebrate tests, survival and
growth in chronic fish and invertebrate tests, and population growth in chronic algal tests as
required in Section 5.2.4 of the G9 (IMO, 2008). Tests are designed with a dilution series to
allow calculation of daily LC50 (concentration yielding 50% lethality) values from acute and
chronic mortality data. In addition, chronic tests will include sufficient treatment replication to
allow calculation of NOEC (no observable effect concentration), LOEC (lowest observable
effect concentration) and EC25 (percent concentration yielding a 25% effect) values for all
toxicity endpoints as required in Section 5.2.5 of the G9 (IMO, 2008). Statistical analyses will
be performed using ToxCalc statistical software (TSS, 2006) according to methods from USEPA
(2002) and ASTM (2006) guidance documents. Briefly, LC50s at daily intervals will be
calculated from survival data using the Probit Method if an adequate dose response is achieved.
If an adequate dose response is not achieved (e.g., only one partial mortality between the
concentration causing 100% mortality and that causing 0% mortality), the Trimmed Spearman-
Karber Method will be used. Chronic data will be tested using a Probit Method (EC25) and by
analysis of variance (ANOVA) with means testing (NOEC/LOEC). Prior to ANOVA testing
chronic data will be tested for normality using the Shapiro-Wilk’s Test and for homogeneity of
variance using the Bartlett’s Test. Survival data will be arcsine square root transformed prior to
analysis. If normally distributed and homogeneous survival and growth data will be analyzed
using a one-tailed ANOVA followed by a Dunnet’s means comparison test (equal number of
replicates/treatment) or a T-Test with Bonferroni Adjustment (unequal replicates/treatment) to
determine differences from control data. If data do not pass the assumptions of normality or
homogeneity, a Steel’s Many-One Rank Test (equal replicates) or Wilcoxon Rank Sum Test with
Bonferroni Adjustment (unequal replicates) will be performed. A
p
value of 0.05 will be used
for all hypothesis tests; a
p
value of 0.01 will be used for testing assumptions of normality and
homogeneity of variance. Results from the chronic statistical analyses will provide NOECs,
LOECs, and EC25s for each ballast water treatment run.
Definition of Test Failure on the Grounds of Toxicity:
Permissible residual toxicity will follow the guidelines outlined by the EPA National
Pollutant Discharge Elimination System (NPDES) for issuance of a Vessel General Permit
(VGP) (full text is available at www.epa.gov/npdes/vessels; relevant sections on ballast
discharge toxicity are 5.8.1.2 and 15.2). Based on these criteria a
test trial will be considered a
failure on the grounds of residual toxicity upon discharge if acute lethality (as indicated by
determination of an LC50 of less than 100%) occurs in any test species. Determination of test
failure as a result of chronic toxicity will be based on EC25 analyses. An EC25 is a point
estimate of the toxicant concentration (expressed as percent effluent) that causes an observable
adverse effect in 25 percent of test organisms. Chronic test results will be calculated in TUc