26
A. MACFADYEN
for arthopod extraction. They do not depend on the animals’ behavioural characteristics and are particularly suitable for the treatment
of deeply worked arable soils containing little plant matter. They also
have the advantage of flexibility of sample-unit size: the same extraction apparatus can be used for treating sample-units of varying
sizes and there is no objection to increasing or decreasing the size
according to statistical requirements as indicated in Section 111.
V. THE CHOICE OF COMPLETE SAMPLING SYSTEMS
FOR SPECIFIC PURPOSES
It has been repeatedly stressed that a sampling system needs to be
planned as a whole. Usually one or two main features of an investigation are irrevocably determined by the purpose of the whole study, for
example that the population size of a particular species is to be quantitatively determined in a limited locality or that the species spectra of
a large number of sites are to be compared and related to botanical and
topographical features of a large area. In particular, the number of
species and of higher taxonomic categories, the exactness of quantitative estimations, the nature of the soil and the range of soil types can
all be studied either intensively or over a broad range. Wherever any
of these requirements give any latitude for choice the experimenter
should, naturally, relax the technical demands as much as possible. If
the groups to be studied are unimportant the easier ones for taxonomy
and extraction (as indicated in Table 11) should be chosen. If the
exact location is unimportant then places with relatively few species
and with uniform light soils will naturally be chosen. If the study is of
the nature of a community survey, time will not be wasted over exact
counts from large samples.
As regards the choice of extraction methods, the advantage of the
different principles have already been discussed under the appropriate
headings, and the applications of particular treatment methods are
briefly summarized in Table 11.
In order to illustrate more specifically what the above precepts imply
three complete sampling schemes will be briefly discussed.
1. An investigation into the effects of different artificial treatments
on the small arthropods of rough grassland.
The statistical aspects of this kind of problem, which is well suited
to demonstrating clearly the effects of simpler environmental factors
are well known. The treatments are applied to a latin square or randomized block design and their effects are determined by the analysis
of variance technique (Snedecor, 1946) or by a partial regression technique (Satchell, 1955). The blocks subject to different treatments are
gridded and regular sample-units taken from each block on a random
A. MACFADYEN
for arthopod extraction. They do not depend on the animals’ behavioural characteristics and are particularly suitable for the treatment
of deeply worked arable soils containing little plant matter. They also
have the advantage of flexibility of sample-unit size: the same extraction apparatus can be used for treating sample-units of varying
sizes and there is no objection to increasing or decreasing the size
according to statistical requirements as indicated in Section 111.
V. THE CHOICE OF COMPLETE SAMPLING SYSTEMS
FOR SPECIFIC PURPOSES
It has been repeatedly stressed that a sampling system needs to be
planned as a whole. Usually one or two main features of an investigation are irrevocably determined by the purpose of the whole study, for
example that the population size of a particular species is to be quantitatively determined in a limited locality or that the species spectra of
a large number of sites are to be compared and related to botanical and
topographical features of a large area. In particular, the number of
species and of higher taxonomic categories, the exactness of quantitative estimations, the nature of the soil and the range of soil types can
all be studied either intensively or over a broad range. Wherever any
of these requirements give any latitude for choice the experimenter
should, naturally, relax the technical demands as much as possible. If
the groups to be studied are unimportant the easier ones for taxonomy
and extraction (as indicated in Table 11) should be chosen. If the
exact location is unimportant then places with relatively few species
and with uniform light soils will naturally be chosen. If the study is of
the nature of a community survey, time will not be wasted over exact
counts from large samples.
As regards the choice of extraction methods, the advantage of the
different principles have already been discussed under the appropriate
headings, and the applications of particular treatment methods are
briefly summarized in Table 11.
In order to illustrate more specifically what the above precepts imply
three complete sampling schemes will be briefly discussed.
1. An investigation into the effects of different artificial treatments
on the small arthropods of rough grassland.
The statistical aspects of this kind of problem, which is well suited
to demonstrating clearly the effects of simpler environmental factors
are well known. The treatments are applied to a latin square or randomized block design and their effects are determined by the analysis
of variance technique (Snedecor, 1946) or by a partial regression technique (Satchell, 1955). The blocks subject to different treatments are
gridded and regular sample-units taken from each block on a random