Читать книгу Practical Field Ecology - C. Philip Wheater - Страница 28

Once a topic for research has been chosen, you can work out the aims of the study. These are important, since tightly defining the aims helps to focus more clearly on the work in hand and can avoid problems in implementation. ‘Woolly’ aims, such as ‘to investigate invertebrates under logs’, may be a starting point for a more focused aim, such as ‘to determine whether the number of invertebrates found under logs is related to the size of the log’. This then leads to further questions, including:

 Which invertebrates are to be examined, i.e. should they be identified to species, or merely counted en masse, or allocated to ecological functional groups (e.g. predators, herbivores, etc.)?

 What is a log (i.e. when is a fallen piece of wood a log rather than a twig?) and how many logs should be investigated?

 How should we standardise or otherwise account for the condition and type of the logs (degree of decomposition, species of tree, etc.)?

 Which measurements of size should be incorporated (e.g. length, width, surface area touching the ground, volume, depth of log in the soil, presence of other organisms such as fungi, etc.)?

 Where should we sample the logs?

 Which statistical method(s) should we use to analyse the data?

Once these questions have been answered, they become objectives that can be used to determine your methods. The aims and objectives lead us to the setting up of working hypotheses. For example, in our study of possible relationships between log size and the numbers of invertebrates found beneath them, we would set up a statistical hypothesis to be tested. It is common practice that the statistical hypothesis to be tested is a null hypothesis; in this context that ‘there is no relationship between log size and the number of invertebrate animals found underneath them’. Most univariate statistical tests examine the likelihood of the null hypothesis being true (see Chapter 5). A null hypothesis should meet the following criteria:

 be a single, clear and testable statement – where more complex research questions are asked, you should break these hypotheses down into individual statements that are treated separately and tested in turn;

 have an outcome, typically either ‘accept’ or ‘reject’ the null hypothesis;

 be readily understandable to someone who is not a scientist.