QSARs for Predicting Cation Toxicity, Bioconcentration, Biosorption, and Binding - Fundamental QSARs for Metal Ions - page 186

between cation concentrations and reproductive impairment. The 2 QSARs relied
on standard reduction-oxidation potential (ΔE 0 ), atomic number (AN), and the
difference of the ionization potential in volts between its oxidation number (OX) and
the next lower one (OX - 1) (ΔIP). Both QSARs for 7 and 12 cations had low standard
deviations and high coefficients of determination ( Table 5.8 ) .
Kaiser (1985) used data from Williams et al. (1982) to develop 3 QSARs that
incorporated ΔE 0 (Table 5.8). Williams et al. (1982) used intraperitoneal injections
in mice to develop the relationship between cation concentrations and mouse 14-day
LD 50 values. The 3 QSARs relied on ΔE 0 , AN, and ΔIP. The most statistically sig-
nificant QSAR was developed for Ba 2+ , Be 2+ , Mg 2+ , and Sr 2+ ; when Y 3+ was added to
these 4 cations, r 2 decreased but SE remained the same (Table 5.8).
Newman's group published a series of papers from 1996 to 1998 describing
quantitative ion-character activity relationships (QICARs) for predicting metal ion
toxicity (McCloskey et al. 1996; Newman and McCloskey 1996; Tatara et al. 1997;
Tatara et al. 1998; Newman et al. 1998). McCloskey et al. (1996) and Newman and
McCloskey (1996) developed 8 and 10 QSARs, respectively, to predict decrease in
bioluminescence of Vibrio fischeri ( Table 5.1 ). This team (McCloskey et al. 1996;
Newman and McCloskey 1996) developed 3 of these QSARs using ΔE 0 ( Table 5.9 ) .
The 3 QSARs relied on ΔE 0 , AN, and ΔIP. QSARs with ΔE 0 and AN/ΔIP had
improved statistical significance compared to QSARs that only used ΔE 0 ; using
the log AN/ΔIP did not improve statistical significance (Table 5.9). Tatara et al.
1997 and Tatara et al. 1998 then developed 9 and 10 QSARs, respectively, to pre-
dict 24-hour LC 50 values for the soil nematode, Caenorhabditis elegans ( Table 5.1).
They developed 2 and 4 of these QSARs using ΔE 0 ( Table 5.10 ). Their two com-
mon QSARs relied on ΔE 0 , AN, and ΔIP. Adding AN/ΔIP to the QSAR with ΔE 0
increased statistical significance for 9 cations, but not for 17 cations (Table 5.10).
Tatara et al. (1998) developed 2 additional QSARs that also used ΔE 0 to predict
24-hour LC 50 values for the soil nematode, Caenorhabditis elegans (Table 5.10).
One used the covalent index X r
m

( ) in addition to ΔE 0 ; the other used the abso-
lute value of logarithm of the first hydrolysis constant (|log K OH |) in addition to
ΔE 0 . Adding |log K OH | to ΔE 0 provided the most statistically significant QSAR
(Table 5.10). Newman et al. (1998) provided a very comprehensive coverage of
these studies and others.
Enache et al. (2000) developed 11 QSARs to predict a 50% reduction of relative
growth of outer leaves in cabbage, Brassica oleracea L. var. capitata (Table 5.1). Of
these 11 QSARs, 9 used ΔE 0 ( Table 5.11 ) . Six of these QSARs used 12 cations and
combinations of ΔE 0 , AN, atomic weight (AW), ΔIP, crystal ionic radius (r), Allred-
Rochow electronegativity (X AR ), Pauling's electronegativity (X), and charge on the
ion (Z). The simplest QSAR with ΔE 0 and both electronegativity properties was
the most statistically significant (Table 5.11). Three of these QSARs used 11 cations
and combinations of ΔE 0 , AN, atomic weight (AW), ΔIP, Allred-Rochow electro-
negativity (X AR ), and Pauling's electronegativity (X) (Table 5.11). Eliminating Cu 2+
produced the most robust QSAR, again with ΔE 0 and both electronegativity properties
(Table 5.11). The outlier position of Cu 2+ is due to its exceptionally high toxicity
determined experimentally by Hara and Sonoda (1979). The test system of Enache
et al. (2000) is listed in Table 5.7 .