THE USE O F STATISTICS IN PHYTOSOCIOLOQY
61
interrelationships can be economically displayed, the investigator may
with greater confidence suggest an hypothesis concerning the causal
factors involved. Such an hypothesis may be regarded as a model
system which would serve to generate the system under examination.
However, if hypothesis-generation is the ultimate aim, it may be
more useful to begin with an a priori hypothesis as to the nature of the
model; for example, it may be postulated that a given number of causal
factors is involved. It may then be possible, first, to test this primary
hypothesis ; and, second, to simplify the data in such a way as to display
the simplest set of interrelationships which would be consistent with
the postulated degree of complexity of the model.
The two approaches are more different than may appear at first sight,
and require quite different statistical techniques. In the one case the
data are unimpaired, the manipulation is directed solely to simplification, and the model arises subsequent to all numerical manipulation.
In the second case, the nature of the model required itself determines the
nature of the manipulation. We find it convenient to distinguish these
two different activities as “primary” and “derivative” model-making.
Both types of method may be used to generate hypotheses; but the
mathematical routes by which the end-products are sought are different,
and the distinction cannot be avoided. It is, moreover, very important
to realize that any system of organizing information to generate rather
than to test hypotheses is not self-sufficient in a scientific context : some
further observations outside the immediate system are required before
such hypotheses can be accepted or rejected.
2. The DeJnition of “Phytosociology”
Whereas plant ecology is usually defined as the study of plants in
relation to their habitat, phytosociology is based on a study of the interrelationships of the plants themselves within an area of vegetation.
Taken at face value, the term “phytosociology” is self-explanatory, in
that it refers merely to the study of plants as gregarious entities. In
this sense, it could equally well apply to the study of the interrelationships of the vegetative offshoots of a single parent, or to a set of
individuals of a single species growing together in the field. More usually,
however, the term is taken to mean the study of sets of species forming
communities under natural or semi-natural conditions. The use of the
word “community” in turn imparts the concept of separate units of
vegetation, comprised of sets of plants with at least a degree of internal
organization in their interrelationships. Moreover, the current identification of phytosociological studies with certain schools of thought has
tended to restrict the meaning of the term still further, so that it is now
CZ
C.E.R.
61
interrelationships can be economically displayed, the investigator may
with greater confidence suggest an hypothesis concerning the causal
factors involved. Such an hypothesis may be regarded as a model
system which would serve to generate the system under examination.
However, if hypothesis-generation is the ultimate aim, it may be
more useful to begin with an a priori hypothesis as to the nature of the
model; for example, it may be postulated that a given number of causal
factors is involved. It may then be possible, first, to test this primary
hypothesis ; and, second, to simplify the data in such a way as to display
the simplest set of interrelationships which would be consistent with
the postulated degree of complexity of the model.
The two approaches are more different than may appear at first sight,
and require quite different statistical techniques. In the one case the
data are unimpaired, the manipulation is directed solely to simplification, and the model arises subsequent to all numerical manipulation.
In the second case, the nature of the model required itself determines the
nature of the manipulation. We find it convenient to distinguish these
two different activities as “primary” and “derivative” model-making.
Both types of method may be used to generate hypotheses; but the
mathematical routes by which the end-products are sought are different,
and the distinction cannot be avoided. It is, moreover, very important
to realize that any system of organizing information to generate rather
than to test hypotheses is not self-sufficient in a scientific context : some
further observations outside the immediate system are required before
such hypotheses can be accepted or rejected.
2. The DeJnition of “Phytosociology”
Whereas plant ecology is usually defined as the study of plants in
relation to their habitat, phytosociology is based on a study of the interrelationships of the plants themselves within an area of vegetation.
Taken at face value, the term “phytosociology” is self-explanatory, in
that it refers merely to the study of plants as gregarious entities. In
this sense, it could equally well apply to the study of the interrelationships of the vegetative offshoots of a single parent, or to a set of
individuals of a single species growing together in the field. More usually,
however, the term is taken to mean the study of sets of species forming
communities under natural or semi-natural conditions. The use of the
word “community” in turn imparts the concept of separate units of
vegetation, comprised of sets of plants with at least a degree of internal
organization in their interrelationships. Moreover, the current identification of phytosociological studies with certain schools of thought has
tended to restrict the meaning of the term still further, so that it is now
CZ
C.E.R.