66
J. M. LAMBERT AND M. B. DALE
eventually relating plant/site and site/habitat systems are still at a very
early stage, and such methods as exist will only be dealt with briefly
in a subsequent section (IIIF). As already implied, the immediate
purpose of the present contribution is to examine methods of phytosociological analysis : it is the plant/site system, therefore, which must
be now more fully explored.
B. THE NATURE O F THE VARIABLES C O N C E R N E D
In any investigation of plant/site relationships within a given region,
the first concern is to define and identify the entities to be recorded.
The sites have already been defined as individual positions on the surface
of the earth: they are thus conveniently identified for record purposes
by any suitable spatial coordinates, such as individual map references.
For independent identification, therefore, the units of plant material
must now be separately defined in terms of characteristics other than
spatial position.
For certain types of vegetation study, it may be appropriate t b record
the variation in plant material from site to site in terms of a single
measurable property. For instance, the variation in total plant mass
from one part of an area to another is often used in studies primarily
concerned with general vegetational productivity ; or differences in
density of an individual species may be required in autecological work.
These are essentially univariate systems, however, and do not concern
us here. I n contrast, in phytosociological work, the plant cover is
assumed to be composed of a number of different units, separately
defined, and each capable of variation across the sites. A phytosociological system is therefore multivariate in nature, and any statistical
analysis of such a system must thus make use of multivariate methods.
Theoretically, plant units could be defined, identified and recorded
by reference to any property it is possible to conceive. There is in
principle no reason, for instance, why properties like chemical composition or geometrical form should not be used, though clearly the
more bizarre the unit, the more restricted will be the application of
results. It is important to realize, however, that, whatever criteria are
used for definition, the choice itself is a subjective step, and depends
on the ultimate purpose for which the information is required. For
particular purposes, therefore, criteria may be selected relevant to that
particular investigation alone. Thus tree-size classes might be required
for the study of tree-regeneration in a given wood; or life-form records
might be appropriate for a study of vegetation on an inter-continental
scale.
Assuming that the ultimate aim of the investigation is to reveal some
structure in the system in order to interpret it in terms of other
J. M. LAMBERT AND M. B. DALE
eventually relating plant/site and site/habitat systems are still at a very
early stage, and such methods as exist will only be dealt with briefly
in a subsequent section (IIIF). As already implied, the immediate
purpose of the present contribution is to examine methods of phytosociological analysis : it is the plant/site system, therefore, which must
be now more fully explored.
B. THE NATURE O F THE VARIABLES C O N C E R N E D
In any investigation of plant/site relationships within a given region,
the first concern is to define and identify the entities to be recorded.
The sites have already been defined as individual positions on the surface
of the earth: they are thus conveniently identified for record purposes
by any suitable spatial coordinates, such as individual map references.
For independent identification, therefore, the units of plant material
must now be separately defined in terms of characteristics other than
spatial position.
For certain types of vegetation study, it may be appropriate t b record
the variation in plant material from site to site in terms of a single
measurable property. For instance, the variation in total plant mass
from one part of an area to another is often used in studies primarily
concerned with general vegetational productivity ; or differences in
density of an individual species may be required in autecological work.
These are essentially univariate systems, however, and do not concern
us here. I n contrast, in phytosociological work, the plant cover is
assumed to be composed of a number of different units, separately
defined, and each capable of variation across the sites. A phytosociological system is therefore multivariate in nature, and any statistical
analysis of such a system must thus make use of multivariate methods.
Theoretically, plant units could be defined, identified and recorded
by reference to any property it is possible to conceive. There is in
principle no reason, for instance, why properties like chemical composition or geometrical form should not be used, though clearly the
more bizarre the unit, the more restricted will be the application of
results. It is important to realize, however, that, whatever criteria are
used for definition, the choice itself is a subjective step, and depends
on the ultimate purpose for which the information is required. For
particular purposes, therefore, criteria may be selected relevant to that
particular investigation alone. Thus tree-size classes might be required
for the study of tree-regeneration in a given wood; or life-form records
might be appropriate for a study of vegetation on an inter-continental
scale.
Assuming that the ultimate aim of the investigation is to reveal some
structure in the system in order to interpret it in terms of other