7765

• UCREFRP Catalog Number: 7765
• Author: Beyers, D. W. and P. J. Sikoski
• Title: Acetylcholinesterase Inhibition in Federally Endangered Colorado Squawfish Exposed to Carbaryl and Malathion
• USFW Year: 1993
• USFW - Doc Type: Environmental Toxicology and Chesistry
• Copyright Material: YES

AChE inhibition in Colorado squawfish

937

Table I. Threshold and effect concentrations for AChE inhibition estimated from 24-h exposure
of Colorado squawfish to technical carbaryl and malathion

Toxicant

Slope

Intercept

Thresholda
(/Lg/L)

7.40 (4.93-Il.l)
150 (83.8-270)

29.3
371

49.1
707

Plateau

Carbaryl
Malathion

0.0419
0.0393

-0.00281
-0.00720

0.0500
0.0914

NOEC
(/Lg/L)

LOEC
(/Lg/L)

....

Plateau, slope, and intercept estimates for the linear-plateau regression equation: AU = intercept + slope [log2(concentration)),
AU (enzyme activity unit) = /Lmol substrate hydrolyzed/mg brain tissue/min.
aThreshold values and 95070 C.l.s (in parentheses) were derived from regression and back-transformed for comparison to NOECs
and LOECs.

..

analysis were 7.40 /-Ig/L (4.93-11.1) carbaryl and 150 ILg/L
(83.8-270) malathion. The 95070 C,I.s for threshold estimates
did not include NOECs, Based on NOECs, carbaryl was ap-
proximately 13 times more toxic than malathion. Relative po-
tency of the two toxicants was confirmed by linear-plateau
regression analysis: Threshold estimates showed that carba-
ryl was about 20 times more toxic than malathion (see Beyers
[13] for summaries of measured toxicant concentrations and
AChE activity).
For each toxicant, the distribution of data as a function of
concentration was such that a linear-plateau regression model
was appropriate (Fig. 1), In both cases, the regression ac-
counted for a significant amount of the total variation
(p = 0,001).

DISCUSSION

Regression analysis vs. ANOVA

Threshold concentrations estimated by linear-plateau re-
gression were lower than NOECs, but the two statistical
methods did not lead to different conclusions. Threshold
concentrations for carbaryl and malathion were approxi-
mately 2.5 and 4.0 times lower, respectively, than the NOECs.

The tendency of linear-plateau regression to produce lower
estimates of effect concentrations may suggest that the pro-
cedure underestimated the concentration at which toxic ef-
fects appeared; however, inspection of Figure 1 shows that
fitted regression models and estimated thresholds were ap-
propriate. A more likely explanation for the discrepancy be-
tween regression and ANOVA effect-concentration estimates
is related to experimental design and analysis. A weakness
of our study was that we used only three replicates per test
concentration, Statistical power would have been increased
had there been more replication, but other disadvantages of
ANOV A (sensitivity to the selected level of significance, sen-
sitivity to variability of data, and dependence on concentra-
tion interval) would not have been rectified. The influence
of these factors has been thoroughly summarized (7,16-18].
Unlike regression analysis, ANOVA does not use concentra-
tion-response relations in toxicological data. ANOVA uses a
pairwise comparison procedure to determine if the difference
between the mean response at a given concentration and the
mean for the control is greater than the minimum statistically
significant difference, In contrast, linear-plateau regression
simultaneously uses all of the data to describe the degree of

5'
~
~ 0.05
:>
i= 0
~ 0,04 0---------0--, ,
* 0,
W &.
U) ,
~ 0,03 @'Q,
w '()
I- , ,
U) - - - - - Carbaryl 'Q
w ,
z 0.02 - Malathion
. :::i
0
:r:
()
..J 0.01
~ Control 1 10 100 1,000 10,000
() TOXICANT CONCENTRATION (ug/L)
<(

Fig. I. Estimated linear-plateau regression lines for brain AChE activity in Colorado squawfish exposed to technical carbaryl and malathion.
Each point represents a mean of three replicates. AU = /Lmol substrate hydrolyzed/mg brain tissue/min. *Trace contamination of solvent
control from air inside vented enclosure.