2466
Environ. Toxicol. Chem. 21, 2002
J.R. Wheeler et al.
availability for the different media and, in particular, the great-
er abundance of uncomplexed or free ionic forms of metals
under freshwater conditions [12,13]. In addition, the greater
ability of saltwater organisms to regulate uptake of zinc and
copper is also likely to contribute to greater tolerance of these
substances by such organisms [14]. It has not been possible
to determine whether or not SSDs tend to coincide when tox-
icity is expressed in terms of the bioavailable form(s) of the
metal because this information is not usually reported. How-
ever, for the purpose of metals risk assessment, this is an
academic point because we can be reasonably confident that
conservative saltwater assessments could be based on fresh-
water toxicity data without a safety factor.
Differences in solubility (and chemical activity) of lipo-
philic compounds between saltwater and freshwater media are
probably due to a salting out phenomenon. Higher salinities
effectively squeeze out neutral organic molecules due to the
strong ionic interactions among water molecules and the major
seawater ions, resulting in reduced solubility in salt water. At
levels below saturation, this means that the effective concen-
tration of the substance is higher, leading to increased activity
and greater bioavailability. This is not a particular feature of
these compounds’ mode of toxic action but rather a conse-
quence of their physicochemical properties. If this is true, then
a greater possibility of risk to saltwater organisms exists, which
will need to be considered in a risk assessment. Alternatively,
one could hypothesize that, as crustaceans (which will often
be as sensitive as insects to insecticides) comprise a dominant
part of the saltwater dataset, they may have introduced a bias
toward greater saltwater sensitivity to the insecticides, espe-
cially if insects and crustaceans are not well represented in
the freshwater dataset. Closer examination of the datasets for
chlorpyrifos, lindane, and malathion shows good representa-
tion in both freshwater and saltwater datasets of insects and/
or crustaceans and, moreover, these were invariably the most
sensitive five or six species of those tested. For these sub-
stances, at least, this explanation is not compelling. However,
for endosulfan, fishes dominated the freshwater dataset (and
the lower tail of the freshwater SSD) while crustaceans oc-
cupied this role in the saltwater distribution. For endosulfan,
apparently greater sensitivity by saltwater organisms may thus
be overestimated due to inadequate representation of crusta-
ceans and/or insects in the freshwater dataset. This could be
investigated by reinforcing the freshwater endosulfan datasets
with crustacean (or insect) data.
perhaps as a consequence of body size [15]). Thus, the most
plausible explanation for this consistent trend is a salting out
phenomenon [16].
A study reported by Zaroogian et al. [17] is useful because
it allows us to examine the ecotoxicological consequences of
such differences in solubility. They developed quantitative
structure–activity relationships for neutral organic compounds
and the mysid shrimp (Americamysis bahia). Toxicity was
consistently underpredicted when predicted toxicities, based
on freshwater solubilities, were compared with experimentally
derived values. For toluene and tetrachloroethylene (where the
difference in solubility was smallest), the solubility adjustment
did not result in a bias toward the experimental data. However,
when solubility data corrected for seawater ionic strength were
applied in the Zaroogian et al. [17] algorithms used to predict
toxicity, the discrepancy was narrowed, at least for six of the
eight chemicals. Despite this, differences between predicted
and measured LC50 values were greater than could be ac-
counted for solely by the influence of salinity on solubility. It
is not possible to identify the precise reason for these differ-
ences, although it is clear that different solubilities in fresh-
water and salt water could go some way toward explaining
differences in SSDs for narcotic compounds.
In practice, a lipophilic test substance will not be in equi-
librium between the test medium and test organisms for much
(possibly all) of the duration of an acute toxicity study. Thus,
it is possible that kinetic factors play a part when the time
taken to reach a critical body burden [18] exceeds the duration
of the toxicity study [19]. It is possible that an increase in
activity of a toxicant (such as would arise due to an increase
in salinity) leads to a greater likelihood of the critical body
burden being reached within the duration of an acute toxicity
study, with consequent differences in the expression of toxicity
in freshwater and saline media. If so, it follows that the dif-
ferences between freshwater and saltwater SSDs reported here
might not be apparent when based on chronic exposure data,
where critical body burdens are more likely to be achieved
within the period of exposure.
In saltwater risk assessments, the greater sensitivity of salt-
water organisms to compounds with a narcotic mode of action
may need to be considered. The difference between freshwater
and saltwater HC5 values was actually rather consistent and
never more than a factor of two (Tables 1 and 2). Therefore,
for risk assessment purposes, only a small safety factor should
be applied when using freshwater data to protect saltwater
species from exposure to narcotic chemicals. However, this
conclusion needs to be qualified in view of small datasets and
especially the lack of data for polar narcotics. It is also note-
worthy that the difference, although small, is a consistent one.
For industrial chemicals, the narcotic mode of action predom-
inates, and so we would advise the generation of additional
saltwater toxicity data for further polar narcotic compounds
to validate this conclusion. Attention to the relationship be-
tween freshwater and saltwater SSDs based on chronic toxicity
data is also warranted.
For saltwater risk assessments, close examination of the
species composition of the freshwater dataset is warranted be-
fore deciding whether or not an additional safety factor needs
to be added to account for possibly greater sensitivity of salt-
water biota. For insecticides, insect and crustacean taxa need
to be well represented, while for herbicides, the presence of
data for algae and higher plants will be key. On the basis of
the data reported in this article, it seems that a modest safety
factor should be applied if freshwater data are used to protect
saltwater organisms from pesticide exposure.
Of the narcotics investigated, only phenol may be regarded
The results for ammonia, metals, pesticides, and narcotics
found in this study can be summarized by ranking the fresh-
water to saltwater HC5 ratios for each substance and plotting
percent ranks against this ratio (Fig. 5). Equivalent sensitivities
are found along the zero line (log scale); those substances
falling to the left indicate greater freshwater sensitivity and to
the right greater saltwater sensitivity. Data for the 21 sub-
as a polar narcotic (log Kow
of nonpolar narcotics (log Kow
ϭ
1.51); the others are examples
2.0–2.54). Again, there was
ϭ
a modest tendency toward greater toxicity to saltwater organ-
isms, although in some cases this conclusion is based on rather
small datasets. Given the nonspecific mode of action of these
compounds, we would not expect a priori any particular tax-
onomic group to be more or less sensitive than another (except
stances closely follow a log-normal distribution (y ϭ 46.47
Lebedew
