THE USE O F STATISTICS I N PHYTOSOCIOLOGY
87
such collation should be attempted over the whole range of sites, and
whole range of species, in the original data, or whether the nodal
abstractions only should be used. Although the former will provide
more information for analysis, the practical difticulties are correspondingly greater. For instance, for a complete analysis of a given area,
habitat observations would be needed for all the sites; but if the
vegetational noda alone are used, a “characteristic” site for each will
have been defined, detailed habitat observations can be concentrated
on these sites, and a great deal of field labour (as well as computation)
w i l l thereby be eliminated. Even the use of noda, however, still requires
other decisions to be made. For instance, it must be decided whether
the nodal aggregates of species are to be represented by their “characteristic” species alone, or whether the other species pertaining to each
nodum are also to be included; further, if the noda are to be used as
separate entities, a decision is needed as to whether they are to be
regarded as of equivalent value on a qualitative basis, or whether they
are to be represented by some form of quantitative measure; and again,
if quantitative measures are required, there is the question of whether
such measures should be based on the status of the noda as regards the
diffuseness or concentration of their composition, on the relative
number of sites occupied by individual noda, or on the number of species
concerned in each case. As in any statistical work, these questions must
be settled by balancing the optimum requirements of the user against
the availability of the appropriate mathematical techniques and the
amount of computing time which would be involved; in short, the
criterion of overall efficiency again is all that can be used in arriving at
the ultimate decisions.
IV. COMPARISON OF PHYTOSOCIOLOGICAL CONCEPTS
A. THE BASIC ASSUMPTIONS
We have deliberately restricted ourselves in the foregoing pages to
the application of statistics to primary survey work, i.e. to situations
where a general assessment of the overall pattern is required before
detailed work can begin on more specific aspects. Here, there are two
apparently irreconcilable views among ecologists as to the best procedure to adopt : one school of thought would apparently wish to try to
fit the vegetation of new areas into a pre-existing classificatory framework and gradually build up a world-wide classification of vegetation
comparable to that of individual plants in orthodox taxonomy; the
other, realizing the magnitude of this task, is content to deal with each
area under investigation as a separate entity and attempt to extract
the maximum ecological information from it. We ourselves believe the
87
such collation should be attempted over the whole range of sites, and
whole range of species, in the original data, or whether the nodal
abstractions only should be used. Although the former will provide
more information for analysis, the practical difticulties are correspondingly greater. For instance, for a complete analysis of a given area,
habitat observations would be needed for all the sites; but if the
vegetational noda alone are used, a “characteristic” site for each will
have been defined, detailed habitat observations can be concentrated
on these sites, and a great deal of field labour (as well as computation)
w i l l thereby be eliminated. Even the use of noda, however, still requires
other decisions to be made. For instance, it must be decided whether
the nodal aggregates of species are to be represented by their “characteristic” species alone, or whether the other species pertaining to each
nodum are also to be included; further, if the noda are to be used as
separate entities, a decision is needed as to whether they are to be
regarded as of equivalent value on a qualitative basis, or whether they
are to be represented by some form of quantitative measure; and again,
if quantitative measures are required, there is the question of whether
such measures should be based on the status of the noda as regards the
diffuseness or concentration of their composition, on the relative
number of sites occupied by individual noda, or on the number of species
concerned in each case. As in any statistical work, these questions must
be settled by balancing the optimum requirements of the user against
the availability of the appropriate mathematical techniques and the
amount of computing time which would be involved; in short, the
criterion of overall efficiency again is all that can be used in arriving at
the ultimate decisions.
IV. COMPARISON OF PHYTOSOCIOLOGICAL CONCEPTS
A. THE BASIC ASSUMPTIONS
We have deliberately restricted ourselves in the foregoing pages to
the application of statistics to primary survey work, i.e. to situations
where a general assessment of the overall pattern is required before
detailed work can begin on more specific aspects. Here, there are two
apparently irreconcilable views among ecologists as to the best procedure to adopt : one school of thought would apparently wish to try to
fit the vegetation of new areas into a pre-existing classificatory framework and gradually build up a world-wide classification of vegetation
comparable to that of individual plants in orthodox taxonomy; the
other, realizing the magnitude of this task, is content to deal with each
area under investigation as a separate entity and attempt to extract
the maximum ecological information from it. We ourselves believe the