THE USE O F STATISTICS IN PHYTOSOCIOLOGY
95
all have their adherents, and there is still no general agreement. If only
qualitative data are available, “dominance” can scarcely have a
meaning, but “constancy” and “fidelity” can still be accommodated.
If the “characteristic species” is extracted statistically by a monothetic
method, it will necessarily be constant in the noda to which it applies;
and if the extraction is made with reference to the parent population
of sites from which each nodum is derived (p. 85), its power of differentiation between noda of closely similar habitats seems likely to invest it
with a reasonable degree of local fidelity. Again, therefore, there seems
to be the possibility of some compromise between opposing schools of
thought.
Since the noda are erected and defined by reference only to the data
from a given situation, they will not necessarily have counterparts in
other areas of vegetation. If the groupings are strong enough, they may
in fact recur in various independent analyses, and we have ourselves
found certain well-marked noda emerging consistently from different
areas analysed statistically; but, although noda may serve as referencepoints for further work on ecological problems within the region
under study, it is doubtful if any very useful purpose would be served
by attempting to use such noda extensively beyond the boundaries
of the general area from which they were derived. However, since the
search for an acceptable method of abstracting, defining and characterizing “vegetation-units’’ is usually associated in ecologists’ minds
with the phytosociological pipe-dream of eventually erecting a series
of such units as the basis for a world-wide classification comparable to
that of orthodox taxonomy, we cannot end this section without at
least some reference to this concept. Whether it is appropriate or
possible to use statistical methods as an aid to such classification, however, seems to us less important than whether any further extension
of existing phytosociological schemes would remain useful for long
enough to warrant the tremendous effort which would be involved.
Vegetation patterns are constantly altering under the increasing
influence of man’s activities, and any general classification would have
to be constantly adjusted. Instead, we should prefer to see the wider
use of more restricted classifications produced quickly and efficiently
in relation to particular regional problems, and discarded without
compunction once their purpose had been served.
V. THE FUTURE OF STATISTICS IN PHYTOSOCIOLOGY
Although probably few modern ecologists would question the value
on theoretical grounds of an objective statistical approach to vegetational problems, the extensive use of statistical methods for large-scale
work is frequently regarded as so impracticable as to be scarcely worth
95
all have their adherents, and there is still no general agreement. If only
qualitative data are available, “dominance” can scarcely have a
meaning, but “constancy” and “fidelity” can still be accommodated.
If the “characteristic species” is extracted statistically by a monothetic
method, it will necessarily be constant in the noda to which it applies;
and if the extraction is made with reference to the parent population
of sites from which each nodum is derived (p. 85), its power of differentiation between noda of closely similar habitats seems likely to invest it
with a reasonable degree of local fidelity. Again, therefore, there seems
to be the possibility of some compromise between opposing schools of
thought.
Since the noda are erected and defined by reference only to the data
from a given situation, they will not necessarily have counterparts in
other areas of vegetation. If the groupings are strong enough, they may
in fact recur in various independent analyses, and we have ourselves
found certain well-marked noda emerging consistently from different
areas analysed statistically; but, although noda may serve as referencepoints for further work on ecological problems within the region
under study, it is doubtful if any very useful purpose would be served
by attempting to use such noda extensively beyond the boundaries
of the general area from which they were derived. However, since the
search for an acceptable method of abstracting, defining and characterizing “vegetation-units’’ is usually associated in ecologists’ minds
with the phytosociological pipe-dream of eventually erecting a series
of such units as the basis for a world-wide classification comparable to
that of orthodox taxonomy, we cannot end this section without at
least some reference to this concept. Whether it is appropriate or
possible to use statistical methods as an aid to such classification, however, seems to us less important than whether any further extension
of existing phytosociological schemes would remain useful for long
enough to warrant the tremendous effort which would be involved.
Vegetation patterns are constantly altering under the increasing
influence of man’s activities, and any general classification would have
to be constantly adjusted. Instead, we should prefer to see the wider
use of more restricted classifications produced quickly and efficiently
in relation to particular regional problems, and discarded without
compunction once their purpose had been served.
V. THE FUTURE OF STATISTICS IN PHYTOSOCIOLOGY
Although probably few modern ecologists would question the value
on theoretical grounds of an objective statistical approach to vegetational problems, the extensive use of statistical methods for large-scale
work is frequently regarded as so impracticable as to be scarcely worth