Till startsida
To content Read more about how we use cookies on gu.se

Transects or plots?

Main conclusion/advice:
• It is possible to use transect data in modelling and mapping, but because of spatial dependence and because it is costly at individual sites, it is often not the best alternative if the main objective is to produce large scale maps

• Sampling in discrete plots has the advantage of being easily standardised. With appropriate considerations about the resolution of sampling and modelling it provides good opportunities for designing precise and cost-efficient sampling programmes.

Transect sampling

The major advantage of transect sampling as a monitoring method is that it covers large areas and therefore it has the potential to provide precise estimates of abundance as well as finding rare species at the particular site of interest. From a modelling perspective, transect sampling also has the advantage of potentially covering a wide range of the important environmental gradients, which improves model performance and generality. Despite these potential advantages, using data from transect sampling in predictive modelling is not straightforward.

Risk for "auto-correlation"
Because transects are so large, they usually need to be split up into smaller units in order to be useful in modeling. For example, a transect of 200 x 5 m might be split up into forty 5 x 5 m units. If video-material is available and if data on positioning is sufficiently accurate, data on vegetation can be collected for individual units. If, however, data was collected from visual estimates during diving, data have to be “invented” for individual units based on large-scale averages. However the data was collected, it is not appropriate to consider all of these as forty good (independent) replicates. This is because samples which are collected close to each other tend to be more similar than those collected further apart (the samples are “autocorrelated”). Taking this into account in the modelling is technically possible, but often complicated in practice. Instead, one common solution is to use only a fraction of the data by selecting samples which are far apart. This means that a lot of resources for sampling may have been spent on collecting data that are not used in modelling and what is most important, this is often done at the expense of number of sites that can be sampled.

Picture?
Continuous sampling in transects is used for collecting information about the cover and presence of vegetation over large areas. A dive transect is typically 10-100 m long and 5-10 m wide while a video transect is 100-1000 m long and 1-5 m wide for video. Data are usually collected by continuous, visual estimates of per cent cover on a semi-quantitative scale or records of individual species.

Plot sampling

Sampling in discrete plots using divers or photographic techniques allows detailed information about occurrence (and abundance) of vegetation in small plots of a controlled size and shape. In general, sampling in plots has the advantage of providing unbiased repeatable estimates of abundance and occurrence with predictable uncertainties. One potential limitation of this method is the small size of the area sampled in an individual plot

Avoid false absences
A “false absence” is when a species is recorded as absent when, in fact, it is present but not found in the sample. The probability of a false absence depends strongly on how rare the species is and what proportion of the area is sampled.
As a rule of thumb it appears that approximately five samples are sufficient for a 80% probability of finding a species (i.e. a 0.2 probability of error), if the frequency of the species in the area is larger than 0.25. To achieve a similar degree of success for species found at a frequency of 0.1, approximately 10 samples are needed.

 

Figure X1. Probability of false absences in 5 x 5 m grid cells using samples of 0.5 x 0.5 m as a function of number of samples (incidentally equal to the per cent area sampled). “p” is the probability of finding the species in one sample. Results are averages from 1000 repeated simulations of sampling in a 5 x 5 m area divided into of 100 0.5 x 0.5 m quadrates.

Alt figurtext: For example, consider a situation where the desired resolution of the model is 5 x 5 m and the quadrate-size is 0.5 x 0.5 m. The probability of false absences can be approximated according to figure X. The figure shows how the probability of error decrease as sample size increase.

Match resolution of samples and model/map
Sampling of (abundance and occurrence of) individual species) in plots are sometimes done by visual estimation but in general it is preferable to use the aid of a grid in the field or on the computer screen to improve the repeatability of sampling. The size of each plot varies depending on the necessary taxonomic resolution but most commonly the area covered by one sample is ≈ 0.1 – 1 m2.

A number of issues arise if only one such sample is taken to represent an area of say 5 x 5 m. In terms of estimating occurrence/distribution (and abundance) of vegetation it may result in uncertain measurements. If sampling is done to support modelling, it will often mean that there is a mismatch between the size of the sampling unit and the grid resolution of the model. This is because the accuracy of positioning systems usually is on the order of ±5 m and because patterns of vegetation at scales smaller than a few meters are influenced by a complex set of unpredictable processes. Therefore, it is likely that for most purposes, models and maps of benthic vegetation cannot have a resolution more detailed than 5 x 5 m. In fact a resolution of 10, 20 or 30 m is probably better.

The solution to the problems of uncertainty and mismatch in resolution is to take several representative samples within an area, which correspond to the resolution of the model! For models of abundance this will reduce the uncertainty of estimated means and for models of occurrence/distribution it will considerably reduce errors due to “false absences”.

 Back to Sampling biological data

© University of Gothenburg, Sweden Box 100, S-405 30 Gothenburg
Phone +46 31-786 0000, About the website

| Map

The University of Gothenburg uses cookies to provide you with the best possible user experience. By continuing on this website, you approve of our use of cookies.  What are cookies?