Method of Studying Biodiversity

Random sampling is usually designed to ensure that each of the population of sampling units has an equal chance of being selected.

Standard statistical methods can then be used to analyze the data. Plot location should not in any way be influenced by any prior knowledge.

Randomly located plots are picked from a numbered list of all plots that could be surveyed, by using random numbers generated by a computer or from tables. Locating plots by eye does not yield randomness, because samples are usually spaced too evenly.

Throwing quadrats to obtain locations, although better than locating by eye, does not achieve true randomness either (this is known as haphazard sampling). Random samples can, however, be time-consuming to locate in the field.

Method of Studying Biodiversity

1. Random Sampling

When the goal of sampling is to provide an indication of what is happening across the whole site, random sampling designs are generally recommended. Random sampling is usually designed to ensure that each of the population of sampling units has an equal chance of being selected.

Standard statistical methods can then be used to analyze the data. Plot location should not in any way be influenced by any prior knowledge.

Randomly located plots are picked from a numbered list of all plots that could be surveyed, by using random numbers generated by a computer or from tables. Locating plots by eye does not yield randomness, because samples are usually spaced too evenly.

Throwing quadrats to obtain locations, although better than locating by eye, does not achieve true randomness either (this is known as haphazard sampling). Random samples can, however, be time-consuming to locate in the field.

Units that are found to fall outside the area are ignored. Transect lines may also be located by utilizing these points. Transects are essentially long, thin quadrats, and many of the same considerations apply. The direction of fixed-length transect lines should usually be randomly allocated.

However, it may be desirable to select a direction that allows samples to be taken along a perceived environmental gradient (e.g. a transition from acid to calcareous grassland). This has the effect of reducing variation between transects, thereby improving precision.

Sometimes it is impossible not to deviate from randomness when sampling, for instance if access to a particular area is not possible. If the inaccessible area is small this may not matter, but if significant bias is possible, the issue should be documented and population estimates may need adjustment.

2. Systematic Sampling

It is often convenient to take samples at regular intervals, for instance at fixed distances along a river.

However, this method creates one main problem: if the sampling interval constantly coincides with a particular regularity in a species or habitat being monitored, the results will be biased.

For example, if you are sampling vegetation at 10 m intervals, and this interval coincides with the raised parts of a hummock–hollow micro topography (perhaps stretching the example!), the vegetation in the hollows (which may be different) will not be sampled.

The results will therefore give a biased picture of the vegetation. Systematic samples are not placed independently of each other (unlike random samples) so, strictly, statistical analysis is not valid.

However, if a large number of samples are taken, systematic samples can usually be treated as random samples without causing substantial problems, unless a systematic bias such as that outlined above occurs.

Sophisticated statistical techniques have been developed for spatial analysis of both systematic and random samples, which enable distribution and density maps to be developed as well as providing alternative estimation methods.

Systematic sampling can be useful because sample sites are relatively easy to select and relocate, and the approach is often more appealing and straightforward to surveyors.

A particular use may be when trying to map both distribution and total abundance of an organism across a study area.

Read Also : Wastewater Treatment Options

The advantages of a regular distribution of sample sites might then outweigh the population estimation disadvantages, for example, if a distribution map based on a regular grid were the objective of the study.

Grid surveys repeated regularly can provide excellent comparative data to identify potential causes and influences of change.

3. Stratified Sampling

Stratified sampling is very commonly used in environmental monitoring as a way of improving the precision of estimates. Very often there is substantial variation across the site in the feature attributes being measured.

This may be due to environmental gradients or differences in management, for example.

In this situation it makes sense to divide the site into sub-units (strata) that relate to the different values of the attributes being monitored (e.g. different densities of a particular species) and sample each sub-unit separately.

Separate estimates are then made for each stratum, which are then combined to provide an estimate for the whole site.

The Advantages include:

An attribute can be estimated with greater precision, provided that the value of the attribute differs substantially between strata and there is more variation in the attribute between strata than within strata uses and influences of change.

Separate estimates can be made for each stratum if these are of interest in their own right.

Stratification slightly reduces the time taken to randomly locate samples.

To maximize the benefits of stratification the site should be subdivided in such a way that it minimizes the within-stratum variability in the attribute being measured (i.e. strata should be as uniform, or homogeneous, as possible).

This normally requires previous survey data or a preliminary survey to be carried out. Alternatively, you can stratify according to known site variations in habitat or ecological factors, which are believed to influence the feature attributes (e.g. a sudden change in soil type).

Although these divisions are not going to be as accurate, as long as there is lower variability within strata this sampling method will provide better results than simple sampling across the whole site.

If the cost of sampling varies, or the within stratum variance in each stratum differs, sampling should be more intensive in the strata in which the costs of sampling are lower or which are more variable.

Sample size should be proportional to the size of the strata if the costs and variances of each stratum are similar, or in the absence of such information.

4. Multi-Stage and Cluster Sampling

In many situations a site may be so large that a high proportion of time is spent travelling between sample sites.

In this instance cluster or multi- stage sampling could be considered as a means of increasing sampling efficiency and in some instances can improve precision for a given sample size.

Multistage sampling is also known as multi-level sampling or subsampling. With multi-stage and cluster sampling a major sample unit is selected, which is divided up into minor units.

Data are then collected from some or all of the minor units. With cluster sampling all the minor units are sampled, but with multi-stage sampling a random or systematic sample of minor units is selected.

In some cases the minor units are themselves sampled (three-stage sampling) but two-stage sampling is the most common technique.

A common example is one in which the major units are transects and the minor units are quadrats along each transect. If all quadrats are sampled this is known as a belt transect.

The main consideration with this technique is that sample units within each major unit are unlikely to be independent of one another since spatial correlation may occur (i.e. sample units are likely to be more similar the closer they are to each other).

Unless the minor units are sufficiently far apart to avoid this, overall precision is likely to be mainly determined by the variation between the major units.

In cluster sampling, the minor units are usually combined and analysis is reduced to simple random sampling of the major units.

This may still be advantageous, compared with simple random sampling of minor units, if there is a significant reduction in the variation between sampling units as these units get larger.

Thus, cluster and multi-stage sampling are likely to be most useful when the area being sampled is relatively uniform at large spatial scales and most of the variance occurs at small spatial scales (but at scales larger than the size of the sample unit).

Transects will be most effective if oriented along a gradient in the attribute being measured. For example, in a study of tree regeneration around woodland, the transects may be oriented away from the woodland, assuming regeneration will decline with distance.

The precision of the overall estimate is primarily affected by the variance between the mean values for major units and, to a lesser extent, by the variance between minor units within each major unit.

Precision is also affected by the number of units sampled at each level. In order to determine the optimum number of major and minor units to sample, some knowledge of the two variances and of the relative cost of sampling at the two stages is required.

This may be obtained through a preliminary survey, or estimated based on available knowledge of the habitat in question.

A preliminary survey may also be designed to investigate the optimal size of the major units as there will be a trade-off between the benefit of having a large sample of major units and increasing their size to reduce between-unit variation. These methods assume that all minor units are of equal size and that each major unit contains the same number of minor units.

5. Stratified systematic unaligned sampling

This is a variation of stratified sampling that combines the advantages of random and systematic sampling. The area to be sampled is first stratified into equally sized blocks (not strata based on habitat characteristics as in stratified random sampling).

Samples are placed in each block by using different x co-ordinates for each column of blocks but the same x co-ordinate within one column, and different y co- ordinates for each row of blocks but the same y co-ordinate within one row.

Read Also : On-site Municipal Wastewaters Treatment

This technique can be an improvement on stratified random sampling because the systematic misalignment is not subject to localized clustering. This technique does not appear to have been widely used.

The time taken to position samples is similar to that for stratified random sampling.

In comparing the techniques discussed, the stratified techniques exhibited greater overall comparative precision than random or systematic techniques, especially at low sample densities with clustered distributions.

In this situation, the sampling strategies ranked in increasing order of relative precision were: random, systematic (regular), stratified random, stratified systematic unaligned.

6. Adaptive sampling

Another approach to consider for features that have very clustered distributions is adaptive sampling. This involves selecting an initial random or systematic sample.

If the target species is found in a given sampling unit, the adjacent sampling units are also included on the basis that there is a good chance these will also contain the species. Potential advantages include:

Although specialized formulae are required for estimation, adaptive sampling can provide better precision for a given amount of effort than simple random sampling;

The method is more satisfying for surveyors, as they do not have to ignore sightings that fall just outside a sampling unit;

A better picture of the species’ spatial distribution is obtained.

One disadvantage is that the sample size cannot be determined in advance; it will depend on what is encountered in the initial sample.

In summary, the methods of sampling varies; Random sampling is usually designed to ensure that each of the population of sampling units has an equal chance of being selected.

In systematic sampling, it is often convenient to take samples at regular intervals, for instance at fixed distances along a river.

However, this method creates one main problem: if the sampling interval constantly coincides with a particular regularity in a species or habitat being monitored, the results will be biased.

The stratified sampling is very commonly used in environmental monitoring as a way of improving the precision of estimates.

Very often there is substantial variation across the site in the feature attributes being measured. This may be due to environmental gradients or differences in management.

However, there are advantages associated with the use of this method. In many situations a site may be so large that a high proportion of time is spent travelling between sample sites.

In this instance cluster or multi- stage sampling could be considered as a means of increasing sampling efficiency and in some instances can improve precision for a given sample size.

Multistage sampling is also known as multi-level sampling or subsampling. The stratified systematic unaligned sampling is a variation of stratified sampling that combines the advantages of random and systematic sampling.

The advantageous method to be considered for features that have clustered distribution is the adaptive sampling. This involves selecting an initial random or systematic sample. If the target species is found in a given sampling unit, the adjacent sampling units are also included on the basis that there is a good chance that these will also contain the species.