Aquatic Toxicology Lab 8Analysis of Short-Term Chronic Test Data

Ceriodaphnia dubia

1. Data Analysisa. Test Acceptability Criteria

i. 80% survival in controlsii. average of 15 neonates per surviving female in controls

b. Endpointsi. Mortalityii. Reproduction

2. Point Estimations - LC50(a)Probit

3. Hypothesis testing - NOEC/LOECa. null hypothesis - no difference

i. when null hypothesis is shown to be false, the alternate hypothesis is assumed to be trueb. alternate hypothesis - a significant difference exits

i. the hypothesis that cannot be statistically tested or proven4. Procedure

a. refer to flowcharts on pages 153 and 159 of lab 7 handoutb. 4 pieces of information needed to select appropriate analysis method

i. concentrations where survival is significantly different from control(1)Fisher’s Exact Test

ii. distribution of the data(1)shapiro-wilk's and chi-square

iii. homogeneity of variance(1)Bartlett's, Hartley's, Cochran's, and Levene's tests

iv. equality of replicatesc. Survival

i. see flowchart on p.153ii. we will not include concentrations which are statistically significantly different from the

controlsiii. we will use Fisher's test to make this determination

d. Reproduction i. see flow chart p.159ii. NOEC/LOEC calculated using the mean number of neonates produced per adult

including dead females(1)Concentrations with mortality significantly greater than controls (calculated using

Fisher’s Exact Test) are not usediii. Depending on the distribution of the data, 4 tests will be used

(1)Parametric i.e. Data are normal and variances are homogeneous(a)equal number of replicates

(i) Dunnett’s Procedure (b)unequal number of replicates

(i) Bonferroni T-test

(2)Nonparametric i.e. data fail normality or homogeneity of variance tests or both(a)equal number of replicates

(i) Steel’s Many-one Rank Test(b)unequal number of replicates

(i) Wilcoxon Rank Sumiv. Minimum Significant Difference (MSD)

(1)The minimum difference in mean reproductive output necessary for the nullhypothesis (no difference) to be rejected.

(2)An important aspect of the data analysis(a)can vary due to changes in variability around control

(3)Parametric test output has a MSD column(4)MSD cannot be calculated for nonparametric tests(5)Important because it normalizes a test result for variability in the controls and allows

comparison of different tests .e. Download Toxstat from the Aquatic Tox Lab web page

i. www.ias.unt.edu/~waller/aquatic_toxicology/laboratoryii. The file we want is toxstat.zipiii. Save toxstat.zip to either the H: drive or a floppy if you have oneiv. Unzip the file as followsv. click the start button

(1)click run(2)click browse(3)locate toxstat.zip file on either your h: drive or your floppy drive (a:) and click it(4)click open(5)click OK(6)this should begin the unzip program(7)save the files to the h: drive or you floppy drive

f. Sample Data Set - Use Data Set 1 at the end of the handout . This is the same data set as inthe lab 7 handout.i. Click on the start buttonii. Click Runiii. Click on browseiv. Find Toxstat on the a: drive in the analysis\toxstat foldersv. Click on OK and the following window appears

vi. Press any key to get to the next screen

vii.The first step in analysis is to determine if there are mortality differences between theexposed organisms and the controls. This is done using Fisher’s Exact Test. Press theF3 Key to get to the screen with the Fisher’s Exact Test option. Navigate through theoptions using the arrow keys and change the options using the space bar.

viii.Change the NO to YES by pressing the space bar to select the Fisher’s Exact Test andthen press the F5 key to run the test. The data will be entered interactively. At the nextscreen (not shown) press E to enter data. The next screen will ask you to enter the data.Follow the prompts and you will have a screen that looks something like this:

ix. After all data are entered the results screens will appear. Note the highlighted statementregarding the statistical relationship between the control and concentration.

x. Press any key to continue until you get to the summary screen that looks like thefollowing. This shows that mortality was significantly different between the 100% andcontrol groups. This is indicated by the asterisk. What does this mean? We don’t usethe 100% group when calculating the NOEC/LOEC

xi. Now that we have tested for mortality differences we need to test for normality andhomogeneity of variance. Toxstat will perform these analyses as well. Press F6 to returnto the main menu. From the main menu press the F4 key to bring up the followingscreen.

xii.Select O to open a new file. Give the data a name, enter the number of groups includingthe control (do not include groups with significant mortality), a group ID(concentration), and finally enter the number of replicates per group. Use the arrow keysto move the highlighted box. To enter or change data, place the highlighter in theappropriate location and enter the data. The entry will appear at the top of the screenuntil you press enter. Once all group ID data are entered, press the F1 key to enterreproduction data.

xiii.The reproduction data entry page appears as follows. Enter the values for the totalnumber of neonates produced by each mother in each group. When you have entered thedata for one group, press the F6 key to move to the next group. If you want to check orchange data in a previous group press the F5 key. Once all data are entered press the F8key. Remember, don’t enter 100%.

xiv.Press w on the next screen in order to write the data to a file. Name the file in order towrite it to disk. Remember these rules! Some of you had trouble with this last time.(1)give it a full path (for example a:\myfolder\mydata.txt)(2)stick to the 8.3 naming convention (eight characters followed by a period and then

three more characters)(3)Remember what you named it - write it down if you need to.

xv.On the next screen (not shown in this handout) select R to return to the main menu.xvi.At the main menu press the F2 key in order to test for normality and heterogeneity of

variance. Select all tests by highlighting NO and changing it to YES using the space bar.

xvii.Press the F5 key to run the tests. There will be a series of 5 screens, one for each test.Look at the last line for a statement regarding statistical significance. If any one of thetests fail for normality or heterogeneity, then you must follow the appropriate path on theflow chart and use the non-parametric method. This data set passes both the tests fornormality and the tests for homogeneity of variance. Using this information and theflowchart on p. 159, determine the most appropriate analysis. Dunnett's test is mostappropriate.

xviii.Press F6 to return to the main menu. Press the F3 key to go to the summary statisticsand tests of means menu. Turn off the Fisher's Exact test. Turn on the list data,summary statistics, and ANOVA - Dunnett's Test options and make sure thatcontrol<treatment is selected.

xix.Press the F5 key to run the analysis. There will be a couple of screens listing the data(not shown) followed by 2 screens of summary statistics on the raw data. Note theCoefficient of Variation on the second screen. This is a measure of reproductivevariability adjusted for the total number of neonates produced.

xx.The next screen contains the output results of the analysis of variance. Note thestatement about the F statistic at the bottom. Remember the null hypothesis is that thereare no differences in mean reproduction between the exposed groups and the control.With this data set we are rejecting the null hypothesis and accepting the alternative thatthere are differences.

xxi.The next screen is a summary of the Dunnett’s procedure to determine which groups aresignificantly different than the control. The groups with the asterisk next to them are

xxii. significantly different than the controls. Hence the NOEC is the highest concentrationsnot different than the controls and the LOEC is the lowest concentration significantlydifferent than the controls.

xxiii.The final screen gives the MSD for this test.

g. Practice on Data Set 2, then analyze the data set from the test completed today. Be sure touse calculated concentration values instead of the percent values.

Test1

Replicate

Concentration 1 2 3 4 5 6 7 8 9 10

Control 27 30 29 31 16 15 18 17 14 27

1.56% 32 35 32 26 18 29 27 16 35 13

3.12% 39 30 33 33 36 33 33 27 38 44

6.25% 27 34 36 34 31 27 33 31 33 31

12.50% 10 13 7 7 7 10 10 16 12 2

25.00% D/0 D/0 D/0 0 D/0 D/0 D/0 0 0 D/0

Test 2

Replicate

Concentration 1 2 3 4 5 6 7 8 9 10

Control 35 39 30 35 35 34 40 35 31 32

6.25% 35 32 30 37 36 32 32 32 34 34

12.5% 32 34 35 31 31 32 33 33 32 33

25% 34 34 19 22 23 24 26 21 21 22

50% 24 D/23 28 D/24 22 22 24 22 D/20 21

100% D/3 D/10 14 D/11 12 D/5 14 14 15 18

Monday Lab Ceriodaphnia dubia Chronic Test Results

Replicate

Concentration 1 2 3 4 5 6 7 8 9 10

Control 29 27 31 33 37 32 26 37 32 27

6.25% 31 29 28 18 25 27 25 34 36 30

12.5% 29 D/5 31 30 32 25 24 28 33 29

25% 27 27 31 24 32 28 23 25 30 31

50% 17 7 14 D 5 15 4 12 D D

100% D D D D D D D D D D

Tuesday Lab Ceriodaphnia dubia Chronic Test Results

Replicate

Concentration 1 2 3 4 5 6 7 8 9 10

Control 25 D/5 31 D D 23 27 NA 25 23

6.25% 24 17 28 NA 31 22 20 26 38 D/6

12.5% D 22 32 25 34 22 28 32 26 NA

25% 16 20 27 23 17 19 25 27 6 13

50% 9 11 D 7 6 3 7 5 11 1

100% D D D D D D D D D D