Page 148 - The Toxicology of Fishes

128 The Toxicology of Fishes

water (C ), and elimination is assumed to be first order with respect to chemical concentration in the
w
fish (C ). Assuming further that the fish can be adequately represented as a single compartment, a mass
f
balance on this system may be written as:
dC dt = k C w − k C f2 (3.123)
f
1
where k and k are uptake and elimination rate constants, respectively, with units of inverse time. By
2
1
assuming that the specific gravity of fish tissue is approximately 1.0, k can also be expressed as a
1
clearance rate, normalized for body weight (e.g., mL water per g tissue per day). The value of k can
1
then be interpreted by comparing it to physiological measures such as ventilation volume.
At steady state, k C = k C . Because, by definition, the BCF equals C /C , rearrangement of this
2
1
w
w
f
f
equation gives the relationship used to estimate the BCF from kinetic data:
BCF = k k 2 (3.124)
1
Elimination rate constants are typically estimated from the slope of log-transformed depuration data.
Uptake rate constants may then be estimated by fitting uptake data to Equation 3.123 using nonlinear
regression methods. Kinetic models have the advantage of being chemical and species specific and do
not require exposure to steady state. The BCF of a chemical may be unreliably estimated by this method
if the duration or extent of sampling of the elimination phase is inadequate or if an inappropriate model
is used (Stehly and Hayton, 1989b).
QSAR Models
Empirically based QSAR models are used to estimate BCFs for untested compounds based on their
physicochemical properties. Most are linear regression models relating the log of the BCF of a compound
to the log of its octanol–water partition coefficient (Lipnick, 1995). Veith et al. (1979), for example,
developed the following relationship from BCFs in fish for chemicals with log K values ranging from
ow
1 to 7:
.
logBCF = (085 × logK ow) − 07 (3.125)
.
QSAR models of this type have generally been developed using data for halogenated organic chemicals
(Mackay, 1982). Implicit in this approach are the assumptions that: (1) an octanol–water system is an
appropriate surrogate for a fish lipid–water system, and (2) bioconcentration results from thermodynam-
ically driven partitioning between water and the lipid phase of the fish (Hansch et al., 1989). QSAR
models can also be fit to lipid-normalized BCF data. Correspondence with the lipid-partitioning assump-
tion is indicated when lipid-normalized log BCF values are approximately equal to log K (Briggs,
ow
1981). Additional QSAR models have been developed based on water solubility and molecular size
descriptors (Hawker, 1990; Isnard and Lambert, 1988; Schuurmann and Klein, 1988).

Bioaccumulation

Bioaccumulation Referenced to Water
Bioaccumulation of contaminant residues in fish may be referenced to the chemical concentration in
water using a bioaccumulation factor (BAF). BAFs are generally normalized to a fish’s lipid content to
permit comparisons among species and between predatory fish and their prey (to assess biomagnification;
see below). For very hydrophobic compounds, adjustments to the total chemical concentration in water
may be attempted as a means of accounting for reductions in bioavailability due to binding to dissolved
and particulate organic carbon (Burkhard et al., 2003). Because the free chemical concentration in water
may be difficult to measure, these adjustments are generally based on empirically derived relationships
(Chin and Gschwend, 1992). The units obtained from this approach are ng/kg lipid divided by ng freely
dissolved contaminant per liter of water.