Stratified sampling is simply the process of identifying areas within an overall habitat, which may be very different from each other and which need to be sampled separately. Each individual area separately sampled within the overall habitat is then called a stratum. The habitat may be fairly uniform, in which case, this is unnecessary. More here

• This is used where the habitat being sampled is fairly uniform.
  Example: The Woodland Project

• To remove observer bias in the selection of samples.

• Where statistical tests are to be used which require randomly collected data.

• Where a large area needs to be covered quickly.

• If time is very limited.

How many samples?

• To show zonation of species along some environmental gradient.
  e.g. down a sea shore, across a woodland edge.                   Example: The Wetland Survey

• Where there is some kind of continuous variation along a line,
  e.g. across the Centre's heathland strips.

• To sample linear habitats, e.g. a roadside verge.

• Where physical conditions demand it, e.g. sampling a vertical rock face, using a rope to climb it.

• Where time is limited. A line transect can be carried out much quicker than a belt transect.

• To visually illustrate how species change along the line (a transect diagram).
  Keys can be chosen to represent individual species. Vegetation height can be drawn in choosing an appropriate scale. The slope of the line can also be measured when carrying out the transect and incorporated into the transect diagram.

• To show species ranges along the line.
  This will generally show only where the species occurs, not how much of it is present. Example

• A belt transect will supply more data than a line transect. It will give data on the abundance of individual species at different points along the line, as well as on their range.

• As well as showing species ranges along the line, a belt transect will also allow bar charts to be constructed showing how the abundance of each individual species changes within its range. Example

• Belt transect data will allow the relative dominance of species along the line to be determined. Example

What interval should be used?

Transects can either be continuous with the whole length of the line being sampled, or samples can be taken at particular points along the line. For example, every metre, or every other metre.

For both line and belt transects, the interval at which samples are taken will depend on the individual habitat, as well as on the time and effort which can be allocated to the survey.

• Too great an interval may mean that many species actually present are not noted, as well as obscuring zonation patterns for lack of observations. More here

• Too small an interval can make the sampling extraordinarily time consuming, as well as yielding more data than is needed. This can cause problems with presenting the data (line transects) as well as sometimes making it harder to see patterns of zonation because of too much 'clutter'.
  More here

• It is important to make sure that the interval chosen does not happen to coincide with some regularly occurring feature of the habitat. For example, if sampling an old field with ridge and furrow systems still obvious, the interval should not be such that all samples are taken on a ridge, or all the samples in a furrow. (Unless, of course, the purpose of the survey is to identify any differences between ridges and furrows!)

The ideal interval will be chosen by balancing the complexity of the individual habitat with the purpose of the survey and the resources available to carry it out.