172
J. HESLOP-HARRISON
that this exposure of latent genotypic variation has nothing to do with
the uniformity of the test environment; it is simply that this environment increases the penetrance of certain genes compared with the native
one. There is another even more subtle way in which the within-sample
phenotypic variance may be increased, even under supposedly highly
uniform controlled envii-onments : this is when genotypes are carried so
far out of their nor& that developmental regulation is disrupted
(Heslop-Harrison, 1959a). Here it is not a question of the exposure of
genetical heterogeneity - the effect may be seen in what is effectively a
pure line - but rather the amplification to a phenotypically obtrusive
scale of developmental “noise”.
Many of the limitations of the simple transplant technique are removed when methods of varied-environment or reciprocal transplanting
are adopted. The classical example of this approach is the work of
Clausen et al. (1940 et sq.) using transplant stations along a coastal plain
to inland mountain transect in California-work which had a prototype, albeit an unsatisfactory one, in the experiments of Gaston Bonnier.
A further technical problem meriting consideration is the selection of
characteristics for observation in a genecological study. Morphological
characters are those most readily investigated ; yet as Turesson pointed
out repeatedly in his pioneer papers, it is the whole reaction of plant to
habitat which is of adaptive importance, physiological responses being
paramount. In a paper on recognizing adaptive variants, Wilkins
(1960a) points out that as the only ultimate test of adaptation is survival
in the given habitat all other kinds of evidence, including that derived
from comparative cultivation, is bound to be circumstantial. The best
form of such circumstantial evidence, he suggests, is that arising from
the study of correlations between measurable habitat factors and plant
characters. Frequently there will be no possible way of assessing the
biological significance of the features measured, and although the existence of correlations may suggest they are adaptive, an element of uncertainty remains since it is always possible - and indeed for many
morphological characteristics, probable - that what is observed is
itself no more than a by-product of the physiological process basically
concerned. Concentration upon morphologica.1 characters may in fact
lead to incorrect conclusions about the “adaptedness” of populations if
the characters themselves are neutral enough in respect to selection to
vary randomly over the area sampled, particularly when the random
variation is taken to indicate the luck of adaptation. Langlet (1963) has
quoted as an example of this kind of reasoning Clausen’s conclusion
(1951) that since no correlation was detectable between latitude of
origin and a small group of morphological features, no ecological clines
&re present in Layia platyglossa.
J. HESLOP-HARRISON
that this exposure of latent genotypic variation has nothing to do with
the uniformity of the test environment; it is simply that this environment increases the penetrance of certain genes compared with the native
one. There is another even more subtle way in which the within-sample
phenotypic variance may be increased, even under supposedly highly
uniform controlled envii-onments : this is when genotypes are carried so
far out of their nor& that developmental regulation is disrupted
(Heslop-Harrison, 1959a). Here it is not a question of the exposure of
genetical heterogeneity - the effect may be seen in what is effectively a
pure line - but rather the amplification to a phenotypically obtrusive
scale of developmental “noise”.
Many of the limitations of the simple transplant technique are removed when methods of varied-environment or reciprocal transplanting
are adopted. The classical example of this approach is the work of
Clausen et al. (1940 et sq.) using transplant stations along a coastal plain
to inland mountain transect in California-work which had a prototype, albeit an unsatisfactory one, in the experiments of Gaston Bonnier.
A further technical problem meriting consideration is the selection of
characteristics for observation in a genecological study. Morphological
characters are those most readily investigated ; yet as Turesson pointed
out repeatedly in his pioneer papers, it is the whole reaction of plant to
habitat which is of adaptive importance, physiological responses being
paramount. In a paper on recognizing adaptive variants, Wilkins
(1960a) points out that as the only ultimate test of adaptation is survival
in the given habitat all other kinds of evidence, including that derived
from comparative cultivation, is bound to be circumstantial. The best
form of such circumstantial evidence, he suggests, is that arising from
the study of correlations between measurable habitat factors and plant
characters. Frequently there will be no possible way of assessing the
biological significance of the features measured, and although the existence of correlations may suggest they are adaptive, an element of uncertainty remains since it is always possible - and indeed for many
morphological characteristics, probable - that what is observed is
itself no more than a by-product of the physiological process basically
concerned. Concentration upon morphologica.1 characters may in fact
lead to incorrect conclusions about the “adaptedness” of populations if
the characters themselves are neutral enough in respect to selection to
vary randomly over the area sampled, particularly when the random
variation is taken to indicate the luck of adaptation. Langlet (1963) has
quoted as an example of this kind of reasoning Clausen’s conclusion
(1951) that since no correlation was detectable between latitude of
origin and a small group of morphological features, no ecological clines
&re present in Layia platyglossa.