CHAPTER 2: EXPERIMENTAL DESIGN

When this study was carried out the student had two objectives. We have included only one.

Draft Aim:

Food acceptability in two species of snail (Helix aspersa and Cepea nemoralis).

Draft Objective:

To examine the acceptability of the leaves of 14 flowering plants as food sources for two species of snail (Helix aspersa and Cepea nemoralis).

What is the statistical population?

In this investigation there are two treatment variables, species of plant and species of snail. The statistical populations are therefore the two snail species and each plant species.

Will you need to sample? By what method(s)? Reflect on whether your sampling method will generate a representative sample. Indicate roughly the number of samples and/or observations you intend to collect and your rationale.

The snails and the plants will need to be sampled. One approach would be to sample the snails and the plants from the same area and make the assumption that they will be a representative sample of each species as a whole. There are three advantages with this approach. The first is practical in that, for example, it is easier to go to one location and sample many snails than go to many locations to sample only one snail at each location. The second advantage is that any environmental variables that may influence the snails will be more consistent within one habitat. Thirdly the selection of plant species as potential food plants to be tested will be easier when considering only one location rather than trying to integrate information from many locations. Therefore, for our design, we have elected to take this approach.

A possible disadvantage is that snails from different locations may preferentially graze on different plants due to variations in either the snails or the plants or both, and we must recognise this as an assumption when presenting our results.

A large number of snails will be required for this investigation therefore, again for practical reasons, all snails encountered within a given habitat will be collected. It will be assumed that the snails that are found will behave in a manner that reflects the species as a whole. (However, it must be remembered that being visible to collectors is part of snail behaviour, and this may be related to eating habits. This is may be a contributor to sampling error.)

In contrast to the snails relatively little plant material is required. Therefore we suggest that when the snails have been collected the geographical centre of the sample distribution be identified and a new leaf taken from one plant nearest to this centre. This method would be used to collect plant material for all species to be tested. Where the leaves on a species were small the next nearest plant was also used to provide the newest leaf etc until enough material was collected.

Step 8 may lead to a revision in the number of samples chosen for the experiment and must always be carried out therefore as part of the planning process.

What is(are) your treatment(s)?

There are two treatment variables - species of snail and species of plant.

What may cause non-treatment variation?

The maturity, health and preconditioning of the snails.

The maturity and 'quality' of the leaf.

Snails tend to be nocturnal.

Snail preference is likely to vary.

Snail behaviour may be influenced by the environment including the presence of other snails or snail trails.

The snails that were collected were deemed to be mature. Maturity in C. nemoralis was judged in relation to shell size and in H .aspersa shell size and the presence of a solid reflected lip on the peristome of the shell indicated maturity.

The snails were kept in tanks for 20 days on a controlled diet before carrying out the treatment.

Prior to the testing the snails were not fed for 24 hours and then placed in individual containers. The testing was carried out at night.

Snail preference may vary so each test was replicated 5 times.

The environment was kept constant across all tests and as similar to natural conditions as possible without introducing additional non-treatment variables

B.1. What type of investigation am I designing?

In this investigation we are starting out with a question so this is an experiment.

B.2. Which type of hypotheses am I testing?

There are three types of hypotheses that you need to choose between. If you are not sure which type of hypotheses you will be testing read the information in B.2.1 - B.2.3 before deciding. For more information about hypotheses and hypothesis testing read Chapter 4.

1. Does the data match an expected ratio?

or

2. Is there an association between two or more variables?

or

3. Do samples come from the same or different populations?

It is not always easy to decide which type of hypothesis you are testing. In this case the student wanted to compare the acceptability index of a particular plant in two species of snail and to compare the acceptability of a number of different plants species. He had no expectation and did not want to examine the association between the snails and plants. He was therefore testing the third type of hypotheses.

B.2.3. Do samples come from the same or different populations?

To test this type of hypothesis you need to decide if the data are likely to be parametric. To tell if your data are likely to be parametric (Normally distributed) you should refer to Box 3.2 in the book and in the Statistical Software section of the Online Resource Centre. When designing an investigation you can only use criterion a to decide whether your data may be parametric and in this case a decision can be made based on this criterion.

Criterion a. Are the data measured on an interval scale which is therefore quantitative and continuous, such as mm and grams?

In this example the answer is NO, the scale of measurement is an acceptability index. This is a based on the ratio of two proportions and is therefore not an interval scale.

B.2.3.2. Non-parametric tests

Choosing a non-parametric test depends on how many treatment variables you are planning to examine, how many categories in each variable, and how many replicates in each category. In this example there are two treatment variables, snails (with two categories) and plants (with 14 categories). The current design has 5 replicates. From the table you can see that the statistical test that is most likely to be appropriate is either the non-parametric two-way ANOVA or the Scheirer - Ray - Hare test.

The criteria for using the non parametric ANOVA (8.5.1.) are that you:

1. Wish to test for differences in population medians.

2. Have two treatment variables each with at least two categories.

3. Have an orthogonal design

4. Have non-parametric data that can be ranked.

5. Can test both general and specific predictions if there are equal numbers of observations in each sample. If there are not equal numbers of observations in each sample only specific predictions can be tested.

All these criteria are met. The design is orthogonal in that every plant species will be tested by both species of snail. The scale though non-parametric is quantitative and can be ranked. The current plan is for 5 replicates of each test and therefore both general and specific hypotheses can be tested.

The criteria for the Scheirer - Ray - Hare test (8.7.1.) are that you:

1. Wish to test for differences in population medians.
2. Have two treatment variables each with at least two categories.
3. The design is orthogonal.
4. Have non-parametric data that can be ranked.

It is clear that these criteria are therefore also met.

(Columns)

H₀: There is no difference between the median acceptability indices between C. nemoralis and H. aspersa across 14 different potential food plants.

H₁: There is a difference between the median acceptability indices between C. nemoralis and H. aspersa across 14 different potential food plants.

(Rows)

H₀: There is no difference between the median acceptability indices between 14 potential food plants when tested using two species of snail.

H₁: There is a difference between the median acceptability indices between 14 potential food plants when tested using two species of snail.

(Interaction)

H₀: There is no interaction between the species of snail and species of potential food plant when observing the median acceptability indices.

H₁: There is an interaction between the species of snail and species of potential food plant when observing the median acceptability indices.

The plant samples engender a response from the snails that is representative of the response of the snail species to these plant species.

There is no undue influence on the snails' behaviour from the test conditions.

The snails were all at a similar level of maturity.

Food acceptability in two species of snail (Helix aspersa and Cepea nemoralis).

Objective:

To examine the acceptability of 14 flowering plants as food sources for two species of snail (Helix aspersa and Cepea nemoralis).

Experimental Design:

The following is a summary of our experimental design. Clearly there are some further details that need to be included such as the names of the plant species being tested, before this is a finalised method.

At least 75 C. nemoralis and 75 H. aspersa were collected from a single location. Given the number of snails required it was decided to collect every snail of these two species found in an area. The geographical centre point of the sample area was noted. The snails were placed in tanks under controlled humid conditions for 20 days and during this time were fed on a mix of broccoli, carrots and cabbage. After this time the snails were placed in individual containers and not fed for 24 hours. One new leaf from each of 14 flowering plants species found in the locality of the snails was collected. The investigator stood in the centre of the snail sampling area and located the nearest plant of each species to be tested. One new leaf was collected from each of these plants. If the leaves were small then the next nearest plant of the species was located and a second leaf collected. The leaf material was used the same day it was collected. Leaf discs were excised from the lamina using a cork borer. A pilot investigation had indicated that C. nemoralis would eat no more than 3 leaf discs in 12 hours and H. aspersa would not consume more than 5 leaf discs. Therefore for C. nemoralis 3 leaf discs of the test plant and 3 discs of lettuce (the alternate plant) were placed in each container with a snail. For H. aspersa 5 leaf discs of the test plant and 5 leaf discs of lettuce were placed in the container. For each combination of snail species / plant species there were 5 independent replicates. The leaf material was placed in the container in the late afternoon as snails are known to be active at night. After 12 hours the remaining leaf material was collected and the total area measured carefully. From this the area eaten was deduced.

An acceptability index (proportion of leaf area of test plant eaten/proportion of lettuce eaten) was calculated and the data analyzed using a Scheirer-Ray-Hare test. The hypotheses being tested were:

H₀: There is no difference between the median acceptability indices between C. nemoralis and H. aspersa across 14 different potential food plants.

H₁: There is a difference between the median acceptability indices between C. nemoralis and H. aspersa across 14 different potential food plants.

H₀: There is no difference between the median acceptability indices between 14 potential food plants when tested using two species of snail.

H₁: There is a difference between the median acceptability indices between 14 potential food plants when tested using two species of snail.

H₀: There is no interaction between the species of snail and species of potential food plant when observing the median acceptability indices.

H₁: There is an interaction between the species of snail and species of potential food plant when observing the median acceptability indices.