Bioavalibility in Soils - Dealing with Contaminated Sites: From Theory Towards Practical Application - page 738

biotic ligand model is important as an approach to modelling bioavailability and is
described below.
The theory underlying the biotic ligand model is that free metal ions react with
binding sites at the organism - pore water interface and the fraction of binding
sites occupied by the metal of interest governs the toxic response. There is compe-
tition for these binding sites between the contaminant of interest and other ions
present such as H + ,Mg 2+ and Ca 2+ . Thus the bioavailability component of the
model comprises determining the free metal ion activity in the pore water and deter-
mining the partition coefficient between the pore water and the binding sites of the
organism (the biotic ligand). Free metal ion activities are either measured directly
with an ion specific electrode (e.g. Steenbergen et al. 2005 ), or calculated using a
chemical speciation program such as WHAM (Windemere Humic Acid Model) VI
(e.g. Thakali et al. 2006a, b ; Tipping 1998 ). The partition coefficients are derived by
experiments in which toxic effects and ion activities are measured with one variable,
e.g., activity of the metal of interest or pH being varied whilst other variables are
kept constant (see De Schamphelaere and Janssen 2002 ). It is important to note that
the above approach is different from simply correlating metal free ion activity with
the metal concentration in the organism of interest. Sometimes such correlations
exist, but not always as competition exists between the metal of interest and other
ions for the exchange sites on the biotic ligand.
Empirical models generally measure the tissue concentration of a metal in the
organism of interest and a host of soil properties, such as pH, bulk metal concen-
tration, pore water metal concentration, dissolved organic carbon, concentration of
Fe and Al oxyhydroxides, concentration of clay minerals, et cetera. Multiple lin-
ear regression techniques are then used to derive a predictive relationship for tissue
concentrations. A large number of such models exist in the literature for earthworms
and were recently reviewed by Nahmani et al. ( 2007b ). The majority of these models
take the form:

a ∗ Log M s + Log M ew ≈ b (16.6)

where:

concentration of metal in the earthworm (mg kg − 1 )

M ew =

concentration of metal in the soil (mg kg − 1 ) M s =

The problem with these models (be they for earthworms or other organisms)
is that they are hardly ever, if at all, validated with independent data sets, largely
due to the lack of appropriate data. Exposure periods, for example, may differ
between experiments, different species of test organisms may have been used, a
variable present in the regression may not have been measured in another study, et
cetera. Thus, their applicability to different sites and soils is always open to ques-
tion. One of the few studies in which relationships were derived and validated with
independent data is that of Sample et al. ( 1999 ). In this study 26 data sets were
used to derive regression equations and six data sets to validate the data. The best
regression coefficients were obtained when metal body burden was regressed against