72
J . M. LAMBERT AND M. B . DALE
decisions to be made as to the form in which the data shall be subjected
to analysis. If the values are not standardized in some way, the more
variable species will tend to dominate the,anaIysis and information of
value may be lost. I n contrast, if the data are adjusted by reduction to
zero mean and unit variance, the balance will be shifted in favour of
the commoner and rarer species. There is much to be said for distortion
of the original data in this way: it in fact provides for the more equitable disposal of the available information and is thus to be recommended
as likely to be in general more efficient.
B. ORDINATION AND CLASSIFICATION
Unless the data are assumed to be homogeneous in the strict statistical
sense, classical statistical techniques are not available for their analysis,
since these are mostly dependent on a particular distribution --the
multivariate normal - for the variables and are sensitive to departures
from this distribution. With basically heterogeneous data usually to
be accommodated, ecologists have reacted to the situation in one of
two ways. One school, accepting the concept of vegetation as composed
of discrete communities, has proposed systems of cZussiJcation ; the
other, leaning towards the idea of vegetation as essentially continuous,
has welcomed schemes of ordination. In classification, the individuals
(i.e. the sites) are arranged in groups, the members of which have
certain properties in common ; in ordination, the individuals are arranged
on axes, with their properties determining their positions. Both types
of method are essentially “structuring” techniques, in that both are
aimed at seeking a simpler structure than that of the original raw data.
Before the two groups of methods can be-compared, the concept of
continuity in vegetation itself needs further examination; it is rarely
defined in any objective way, and it is by no means clear from the
literature exactly what is intended by the use of terms like “vegetational
continuum”. The position is further confused by the fact that some
authors appear to use the expression in a strictly phytosociological
sense (i.e. independent of the actual position of the sites), while others
equate it with spatial continuity on the gr0und.t
Statistically, continuous variation means that the variables concerned can take any value within the limits of truncation adopted;
strictly, therefore, the concept of continuity can only apply to vegetation with no qualitative differences. In vegetation differing floristically.
therefore, the term needs a wider connotation before it can be useful.
f Since this paper was written, a useful further contribution by Goodall (1963) has
appeared, i n which (pp. 308-310) the distinct concepts of continuity in the “vegetational space” of a geometrical model, and continuity of vegetation i n real space, are
fuuy appreciated and discussed.
J . M. LAMBERT AND M. B . DALE
decisions to be made as to the form in which the data shall be subjected
to analysis. If the values are not standardized in some way, the more
variable species will tend to dominate the,anaIysis and information of
value may be lost. I n contrast, if the data are adjusted by reduction to
zero mean and unit variance, the balance will be shifted in favour of
the commoner and rarer species. There is much to be said for distortion
of the original data in this way: it in fact provides for the more equitable disposal of the available information and is thus to be recommended
as likely to be in general more efficient.
B. ORDINATION AND CLASSIFICATION
Unless the data are assumed to be homogeneous in the strict statistical
sense, classical statistical techniques are not available for their analysis,
since these are mostly dependent on a particular distribution --the
multivariate normal - for the variables and are sensitive to departures
from this distribution. With basically heterogeneous data usually to
be accommodated, ecologists have reacted to the situation in one of
two ways. One school, accepting the concept of vegetation as composed
of discrete communities, has proposed systems of cZussiJcation ; the
other, leaning towards the idea of vegetation as essentially continuous,
has welcomed schemes of ordination. In classification, the individuals
(i.e. the sites) are arranged in groups, the members of which have
certain properties in common ; in ordination, the individuals are arranged
on axes, with their properties determining their positions. Both types
of method are essentially “structuring” techniques, in that both are
aimed at seeking a simpler structure than that of the original raw data.
Before the two groups of methods can be-compared, the concept of
continuity in vegetation itself needs further examination; it is rarely
defined in any objective way, and it is by no means clear from the
literature exactly what is intended by the use of terms like “vegetational
continuum”. The position is further confused by the fact that some
authors appear to use the expression in a strictly phytosociological
sense (i.e. independent of the actual position of the sites), while others
equate it with spatial continuity on the gr0und.t
Statistically, continuous variation means that the variables concerned can take any value within the limits of truncation adopted;
strictly, therefore, the concept of continuity can only apply to vegetation with no qualitative differences. In vegetation differing floristically.
therefore, the term needs a wider connotation before it can be useful.
f Since this paper was written, a useful further contribution by Goodall (1963) has
appeared, i n which (pp. 308-310) the distinct concepts of continuity in the “vegetational space” of a geometrical model, and continuity of vegetation i n real space, are
fuuy appreciated and discussed.