Sampling Strategies and Measurements - "Where and How Many?"
Here we provide the details of why you might choose a particular sampling strategy and a particular measurement strategy.
According to Adds et al. (1999)[1] the purpose of sampling is to obtain a series of independent estimates of the variables you are measuring. If you want to study Zug Lake and the algae population, you cannot possible measure the whole of Zug Lake. You therefore need to take a sample of the area and the biotic and abiotic conditions in that area. Usually some form of random sampling is the best way to select an unbiased sample in a relatively uniform habitat.
"Sampling strategies may be used to measure biotic and abiotic factors and their change in space, along an environmental gradient, over time, through succession, or before and after a human impact (for example, as part of an EIA)." IB ESS Guide 2015
"Measurements should be repeated to increase reliability of data. The number of repetitions required depends on the factor being measured." IB ESS Guide 2015
"Methods for estimating the abundance of non-motile organisms include the use of quadrats for making actual counts, measuring population density, percentage cover and percentage frequency." IB ESS Guide 2015
Random Sampling:
Random sampling of smaller quadrats within a larger defined area (quadrat).
Random co-ordinates can be used to locate sampling positions in a habitat, such as a meadow, woodland or lake. Once the sampling area has been designated then a further sampling area should be defined at the site and marked out using tape measures. The size of the sampling area will depend on the type of study being undertaken.
A quadrat can be any size and so it is important to define what size quadrat you are using so that it is repeatable. For sampling trees, you might use a 100 m2 quadrat, while for lichens you might use a 10 cm2 quadrat. For our plant sampling we will be using 50 cm2 grid quadrats.
Once you have defined your study area, you need to divide the area up into a grid. You then can sample the grid using randomly generated grid coordinates (x,y). This removes the bias of choosing areas that might be interesting or “easy”. It also removes the bias that can be introduced if a quadrat is thrown. True randomness is an important element in ecology, because statistics are widely used to process the results of sampling. Many of the common statistical techniques used are only valid on data that is truly randomly collected. Random samples can be achieved by using a random numbers table or a random number generator, e.g. http://www.mathgoodies.com/calculators/random_no_custom.html.
Within habitats such as woodlands or scrub areas, it is also often not possible to physically lay quadrat frames down, because tree trunks and shrubs get in the way. In this case, an area the same size as the quadrat has to be measured out instead and the corners marked to indicate the quadrat area to be sampled.
How to calculate a representative sample for measurements:
Use a running mean for occurrence of the species of interest.
When investigating biodiversity, use a cumulative frequency chart:
If you are studying an environmental gradient, using transects, then for ESS we say that you must complete at least 3 parallel transects. This is based on the Chief Examiner’s annual report.
If applicable, you should consider what statistical test you are going to apply to your data. Are you looking for a correlation between two factors or a difference between two sets of data? We can use a statistical decision chart to help with this process.
When possible, means and standard deviations (descriptive statistics) should be calculated at the very least. Remember that five repeat measurements are needed for standard deviation to be valid.
Methods for estimating the abundance of non-motile organisms include the use of quadrats for making actual counts, measuring population density, percentage cover and percentage frequency.
Non-Motile Organisms
In order to determine the number of non-motile (usually plants) organisms we have several strategies. This will always include some kind of sampling strategy.
Useful reference: Methods for Plant Sampling
With this method, we can do a number of different measures. We can:
- make actual counts of individuals
- measure or estimate the population density
- measure or estimate the percentage cover
- measure or estimate the percentage frequency
Actual Counts
This is only possible when the species in question has a relatively low abundance otherwise it is prone to error and time consuming. Examples of when this would be suitable are tree counts or rare species. You need to use this method for studies of biodiversity as absolute numbers are needed (abundance) along with identification of species present.
Estimating Population Density
This is possible for plants that grow as discrete individuals. An example would be when you are looking at woodland regeneration. Count the number of oak seedlings[2] in a series of 10 m2 quadrats. The number per unit area is the density of seedlings.
Subjective Estimates of Percentage Cover
It is possible to give a quick estimate of percentage cover. This is useful as a descriptive difference between two areas in support of more quantitative data collection. It is prone to observer error as no two people will give the same estimate generally.
It is also possible to use the Domin Scale. This defines bands of percentage cover and allocates a score. This can add consistency to estimates of cover.
Objective Estimate of Percentage Cover
A gridded quadrat (usually 100 squares) can be used to determine local frequency – the number of squares in which a species is found. These values will depend on quadrat and grid size and so it is essential to state these in the description of the methods used.
Species often overlap, and there may be several different vertical layers. Percentage cover may therefore add up to well over 100% for an individual quadrat. This method can cause an overestimation of the importance of large plants.
Objective Estimate of Percentage Frequency
This is the percentage probability of a plant occurring within a quadrat. To determine frequency you need to record presence and absence data. Presence should be defined. It could be:
- Plants rooted in quadrat
- Plants touching or overhanging quadrat
- Plants which are in or touch the top and right sides of the frame.
Frequency depends on the size of quadrat used. Percentage frequency measures will decrease as the quadrat size decreases.
%Frequency = (no. of quadrats containing plant / total no. of quadrats examined) x 100
Percentage frequency tells us how common a species is, but it does not give information on how much of the species is present once we have recorded it.
Relationship between Cover and Frequency
Percentage cover and percentage frequency are two measures that are obviously related (the more space a species occupies, the more likely it is to be found in more than one sample), but using the two together can give information on how species are distributed in a habitat. For example, if a species was found to have a low % frequency, but a relatively high mean % cover, it would indicate that the species was occurring (rarely) but in large clumps where it did occur. This would suggest that the species had a very clustered distribution. Equally, a species might have a very high % frequency, showing that it was widespread over the habitat, combined with a low mean % cover, which would tell us that while it was common, it was not very dominant in the habitat in terms of % cover.
Here are a couple of supporting sites on sampling techniques for further reading and clarification.
Countryside Info Ecological Sampling Methods
Royal Geographical Society Sampling Techniques
Example Question: May 2001 Paper
1. A population of non-motile (unable to move from place to place) organisms was sampled using quadrats. Each quadrat was divided into one hundred smaller squares. In each quadrat, the percentage frequency was recorded, added to the data obtained from previous quadrats and the new mean was calculated. This is the cumulative mean percentage frequency. The results of this investigation are shown in the graph below.
(b) What is meant by percentage frequency?
(c) Explain how many samples are needed?
(d) What would you expect for the frequency chart if the plant was rare or becoming rare?
Measuring Abiotic Factors
It’s really important that you describe your method of recording abiotic data. It must be repeatable and show you have taken into account the variability of the data. You should name the instrument used to measure the factor and make a note of the built-in error in the instrument, although this is unlikely to be an important factor in ecological investigations.
Here is an example.
Insert Vernier soil temperature probe into the soil to a depth of 10 cm. Wait 30 seconds then record the measurement. Repeat this measurement for a total of 5 times in order to be able to calculate a mean.
Typical abiotic factors
Students should know how to describe the measurement of at least three different abiotic factors from different ecosystem types. Ideally, students should have experience of doing these measurements.
| Terrestrial Ecosystems | Stream and River Ecosystems | Lake / Marine Ecosystems |
| Insolation (light) | Dissolved phosphate/nitrate | turbidity |
| Air temperature | Dissolved oxygen | chlorophyll a |
| Soil temperature | stream flow velocity | dissolved oxygen |
| Soil humidity | water temperature | water temperature |
| Wind speed | pH of water | pH water |
| Air humidity | depth | depth |
| Salinity of soil | wave action | |
| Nitrate / Phosphate content | ||
| pH of soil | ||
| slope / aspect of slope | ||
| altitude | ||
| soil characteristics |
Footnotes
1. Adds, J., Larkcom, E., Miller, R., & Sutton, R. (1999). Tools, Techniques and Assessment in Biology. A Course Guide for Students and Teachers. Nelson Thornes Ltd., Cheltenham.
2. http://www.countrysideinfo.co.uk/plant.htm