FORTY YEARS O F GENECOLOQY
169
In most genecological studies of the last forty years the intention has
been to detect habitat-correlated, genetically-based variation, using
population sampling methods of different kinds followed by comparative cultivation. Many have been open to criticism because the techniques adopted have hardly been adequate to permit the critical distinction between random and non-random variation. As Harberd (1957,
1958) and Wilkins (1959, 1960a) have pointed out in valuable critiques
of genecological methodology, it may be extremely difficult - and in
some cases, indeed, perhaps impossible - to make this distinction by
the orthodox type of genecological trial.
Since the whole object of such trials is to seek for differences between
populations and relate these to habitat, comparisons between populations taken singly and in groups are constantly required. Unless these
comparisons can be placed upon a statistically and biologically sound
basis valid conclusions cannot be drawn. In statistical terms, the
exercise is to partition the genetical variance of the species in such a way
that that part attributable to habitat type can be estimated and its
significance evaluated. The ideal situation for analysis would be one
where a species had encountered a mosaic of habitats while expanding
its range, the selective influences in each habitat acting in turn upon the
whole available pool of genetical variation. A comparison of betweenhabitat and within-habitat variances could then be expected to expose
as significant those differences which were truly adaptive. All practical
cases differ from the ideal one in some respect or other, and some so
radically that analysis if not impossible can yield only dubious results.
The example of regional subspecies illustrates the point. It may be
possible to show that the area of one coincides with a climatic zone
different from that of another; each can certainly be said to be “adapted”
to its environment to the extent that it survives, but it cannot be said
that all the morphological and physiological differences between them
have necessarily been the outcome of differential selection in the two
areas. As Wilkins (1959) has pointed out, the area of one may have been
colonized by a small somewhat unrepresentative invasion from the
other, so that adaptive differences rising secondarily in consequence of
selection are confounded with the original chance differences between
the colonists and the population from which they emerged. In this case
evidence of genecological differentiation must be sought by the experimental demonstration of adaptation, using reciprocal transplant
methods to test survival capacity or controlled environments to measure
response to individual habitat factors.
This difficulty is part of a general one, arising from what Wilkins
(1959) has termed “a general uncertainty about the variance found
within a local population”. Another aspect of this has been stressed by
169
In most genecological studies of the last forty years the intention has
been to detect habitat-correlated, genetically-based variation, using
population sampling methods of different kinds followed by comparative cultivation. Many have been open to criticism because the techniques adopted have hardly been adequate to permit the critical distinction between random and non-random variation. As Harberd (1957,
1958) and Wilkins (1959, 1960a) have pointed out in valuable critiques
of genecological methodology, it may be extremely difficult - and in
some cases, indeed, perhaps impossible - to make this distinction by
the orthodox type of genecological trial.
Since the whole object of such trials is to seek for differences between
populations and relate these to habitat, comparisons between populations taken singly and in groups are constantly required. Unless these
comparisons can be placed upon a statistically and biologically sound
basis valid conclusions cannot be drawn. In statistical terms, the
exercise is to partition the genetical variance of the species in such a way
that that part attributable to habitat type can be estimated and its
significance evaluated. The ideal situation for analysis would be one
where a species had encountered a mosaic of habitats while expanding
its range, the selective influences in each habitat acting in turn upon the
whole available pool of genetical variation. A comparison of betweenhabitat and within-habitat variances could then be expected to expose
as significant those differences which were truly adaptive. All practical
cases differ from the ideal one in some respect or other, and some so
radically that analysis if not impossible can yield only dubious results.
The example of regional subspecies illustrates the point. It may be
possible to show that the area of one coincides with a climatic zone
different from that of another; each can certainly be said to be “adapted”
to its environment to the extent that it survives, but it cannot be said
that all the morphological and physiological differences between them
have necessarily been the outcome of differential selection in the two
areas. As Wilkins (1959) has pointed out, the area of one may have been
colonized by a small somewhat unrepresentative invasion from the
other, so that adaptive differences rising secondarily in consequence of
selection are confounded with the original chance differences between
the colonists and the population from which they emerged. In this case
evidence of genecological differentiation must be sought by the experimental demonstration of adaptation, using reciprocal transplant
methods to test survival capacity or controlled environments to measure
response to individual habitat factors.
This difficulty is part of a general one, arising from what Wilkins
(1959) has termed “a general uncertainty about the variance found
within a local population”. Another aspect of this has been stressed by