Geoscience Reference
In-Depth Information
Percentage scores and other 'closed' measures
Whenever an abundance score is expressed as a proportion or a
percentage any variation in the abundance of one taxon will affect
the relative abundance of other taxa, even though their absolute
occurrences may remain constant between samples. A rare taxon
(few absolute occurrences throughout the population of samples)
may have a high percentage score in a sample where it was the
only taxon present, even though its absolute abundance within
that sample is similar to that in other samples where it
occurred with other taxa. Any closed measurement system,
such as percentage scores, has to be used with great care.
5
Logarithmic transforms and the reduction of 'noise'
Adjacent samples often exhibit great variations in fossil
abundance. This can be produced by the use of an inappropriate
sample size, but such a pattern does seem to be a general
characteristic of most allochthonous assemblages due to 'random'
perturbations in fl uid fl ow associated with the depositional
process. This randomness also produces high levels of statistical
'noise' in the data that can swamp a palaeoecologically useful
pattern. One way of overcoming this is to use a logarithmic
transform. Abundant taxa will exhibit a greater variation in
density than rarer taxa, and such variation can swamp any
palaeoecologically useful pattern. By applying a log transform the
pattern displayed by variations in the rarer components is retained
while the noise associated with the less signifi cant variations in
abundant species is minimized (see Worked Example 5.2, p. 97).
Wrap delicate fossils in tissue
paper (not cotton wool) and
then place in a protective
box. Multiple specimens and
boxes should be tightly
packed in strong containers
to prevent movement and
abrasion during shipping. If
possible try to keep the
container weight to less than
10 kg for ease of handling.
5.5.3 How many samples are required?
It is often diffi cult to decide when suffi cient samples have been
scored to enable adequate investigation of a given assemblage.
Certainly no generalizations can be made as every fossil
assemblage is different. A guide to sample size may be obtained
by constructing a graph in which the cumulative mean (of
whatever abundance measure is being used for a given taxon)
is plotted against sample number (Figure 5.15).
40
30
Figure 5.15
As the number of
samples taken rises so the
fl uctuation in the value of the
cumulative mean of the number of
taxa will decrease. The point at
which the variation stabilizes (in
this case between 25 and 30)
indicates the minimum number of
samples required to investigate that
population.
20
10
5
10
15
20
25
30
35
40
number of quadrats
