7764 - Laserfiche WebLink

Home Browse Search

Toxicity of carbaryl and malathion 103 bonytail. Initial weights and lengths were determined by mea- suring 20 fish sacrificed and preserved at the start of the ex- posure period. Renewal-acute tests with the three toxicants were conducted simultaneously. Ten larvae were placed in replicate exposure chambers for a total of 20 per test concen- tration. Obviously deformed or abnormal larvae were not se- lected. Larvae were not fed within 24 h of the start of a renewal-acute test or during the 4-d exposure period. Survival and behavior were monitored at least daily. At conclusions of renewal-acute tests, surviving fish were sacrificed by ad- ministering an overdose of MS-222 (Argent Chemical Lab- oratories, Redmond, WA) and preserved in 10070 formalin. ELS tests with Colorado squawfish and bonytail were ini- tiated with 41- and 48-d-old larvae, respectively. Mean wet weight and total length, respectively, at the start of tests were 9 mg and 12 mm for Colorado squawfish and 4 mg and 8.6 mm for bony tail. Thirty Colorado squawfish and 40 bonytaillarvae were placed in replicate exposure chambers. Larvae were acclimated to conditions within exposure cham- bers for 48 h before the toxicant-metering system was acti- vated. Larvae were fed live :5 24-h-old brine shrimp nauplii two or three times a day. Approximately 100 nauplii per fish per feeding were introduced into exposure chambers, and the number of nauplii was adjusted to account for mortal- ity of test animals. Larvae were not fed within 24 h of con- clusion of a test. Exposure chambers were siphoned as required to remove debris. Survival and behavior were ob- served daily; however, small size and rapid deterioration of dead larvae made accurate counting difficult. Therefore, counts of fish surviving at the conclusion of an exposure pe- riod were used to estimate survival. Upon conclusion of a test, fish were sacrificed by administering an overdose of MS- 222 and preserved in 10% formalin. Preserved fish were blot- ted, counted, and weighed (:t I mg). Statistical analysis Median lethal concentrations for mortality in renewal- acute tests were estimated by pro bit analysis [15]. Toxicity of carbaryl was compared to that of Sevin-4-Oil by calculat- ing a ratio of the median lethal concentrations estimated for each toxicant. A ratio> 1.0 ratio suggested that Sevin-4-0il was more toxic than carbaryl; a ratio < 1.0 ratio, that Sevin- 4-Oil was less toxic [6]. Ratios were based on a.i. Two methods of analysis - hypothesis testing and regres- sion analysis- were used to analyze survival and growth (as weight) in 32-d ELS tests. For hypothesis testing, survival data were analyzed twice so that effects of two alternative statistical transformations could be assessed. First, a modi- fication of the angular transformation [16] p' = ..} (n + 0.5) arcsin..} (x + 0.375)/(n + 0.75) ~ where n = the initial number of animals in a replicate and x = the number surviving, was applied to stabilize the vari- ance (n was constant in our experiments). Alternatively, the logistic transformation logit = In[(np + 0.5)/(nq + 0.5)] where p = the proportion surviving and q = 1 - p, was ap- plied to survival data [17]. Statistical weights (w) for logis- tic-transformed survival values were calculated with the formula w = Il/(np + 0.5) + l/(nq + 0.5)]-\ Survival and growth of fish in solvent controls and dilu- tion-water controls were compared by calculating a t statis- tic and comparing it to a two-tailed Student's critical value. If effects of the solvent and dilution-water controls were not significantly different (alpha = 0.05 for all statistical com- parisons), data from these two treatments were pooled for subsequent analyses. After pooling control treatments, all data were subjected to Shapiro- Wilk's test for normality and Bartlett's test for homogeneity of variance [16]. No addi- tional transformations were required to meet assumptions of normality or homogeneity of variance. Following formal test- ing of statistical assumptions, angular-transformed survival data were subjected to one-way ANOYA. Logistic-trans- formed survival and growth data were subjected to weighted one-way ANOYA. Weighting factors for growth data were equal to the number of fish in the sample from each repli- cate. Treatments that had significantly different effects com- pared to controls were identified by calculating a t statistic for comparison to a one-tailed Dunnett's critical value. NOECs and LOECs were estimated for survival and growth. For regression analysis, a linear-plateau regression model, also called hockey stick [18] or threshold model [19], was fit to survival and growth data as a function of toxicant con- centration. An assumption for use of this model was that there be a toxicant concentration (threshold) below which toxic effects were not exhibited and above which a concen- tration response was observed. Of particular interest in this analysis was estimation of the threshold and its C,I., because they represented the toxicant concentration at which effects began to be manifested. The linear-plateau regression model had the form {{30 + {3\XO for X:5 Xo y= {3o + {3jX for x ~ Xo where x represented toxicant concentration and Xo repre- sented the threshold concentration. Before regression analysis, survival data were subjected to the logistic transformation. Measured toxicant concentra- tions were log2 transformed. No other transformations of data were made. As in hypothesis testing, survival and growth data were analyzed using weighted analyses. The multivari- ate-secant, nonlinear regression method was used to simul- taneously (a) fit a line through data that composed the plateau (zero slope), (b) fit a second line through data that showed a concentration response (nonzero slope), and (c) es- timate the threshold concentration. Data and residual plots were examined to confirm that regression models were ap- propriate and statistical assumptions were not violated. All statistical analyses were conducted with SAS@ statistical soft- ware [20].