Sie sind auf Seite 1von 11

Transactions of the Royal Society of South Australia (2011), 135(2): 113123

INTERACTION BETWEEN GENOTYPE AND ENVIRONMENT:


A TALE OF TWO CONCEPTS1
OLIVER MAYO
CSIRO Livestock Industries
PO Box 10041 Adelaide BC
South Australia 5066
oliver.mayo@csiro.au

Abstract
Fishers (1918) linear model for quantitative inheritance has been extraordinarily successful, even
though it was attacked, correctly, by Hogben (1933) as unsuitable for the analysis of traits
manifesting non-linear interactions, which were likely to be common, not rare. This prediction
has been confirmed. The robustness and success of the linear model, especially for short-term
prediction in applied genetics, should not be allowed to obscure its difficulties when multivariate
methods can replace it and the data on which to use them are available.
KEY WORDS: Fisher, G E interaction, Hogben, linear model.
NURTURE AND NATURE. Man is so educable an animal that it is difficult to distinguish between than part of
his character which has been acquired through education and circumstance, and that which was in the original grain
of his constitution. Different aspects of the multifarious character of man respond to different calls from without,
so that the same individual, and much more, the same race, may behave very differently at different epochs.
(Galton, 1883)
The progress of research has gone steadily to show that facts of heredity which at first seemed to be hopelessly
complicated can be represented in terms of a strict Mendelian system. This simplification of the problem has far
exceeded our earlier anticipations, and I have to regret that in dealing with several sets of phenomena I
countenanced non-Mendelian interpretations which in almost every case it has been found possible to replace by
simple Mendelian formulae. Where this reduction to a common plan has not been yet effected, the difficulty, we
feel fairly confident, is generally created rather by the disturbance of environmental causes or by the influence of
undetermined factors than by any profound aberrations in physiology.
(Bateson 1909 pp. 245246)

Introduction
Tabery (2008) has reviewed the history of G E interaction from the point of view that Hogben (1933) and
Fisher (1918) took opposite sides on the extent of G E and its importance in the determination of major
human traits such as mental ability. The topic has a much longer history, as exemplified by the epigraphs
above. Furthermore, recent discussion of the topic has included remarkable statements such as The use of
heritability in human genetics follows from its successful application in animal breeding (Guo 1999 p. 215).
This was in the course of an analysis of models of correlated environmental effects that yield non-zero
heritability estimates in the absence of genetical influences on a trait. It therefore seems worthwhile to
consider the topic in further depth.
Definition of interaction
Interaction between two entities simply means that the entities influence each other, that a change in one, from
whatever cause, induces some change in the other, or in an outcome in which the entities share. In the case
of interaction between genotype and environment, the outcome is the phenotype, and this interaction then
simply refers to the process whereby genotype and environment jointly produce a phenotype (Table 1).
1. Test of a paper presented to a meeting of the Royal Society of South Australia on Thursday 10 March 2011.

113

O. MAYO
Table 1. Classification of genotype environment interaction.
Type

Example

Analysis

Allele another particular allele of a gene

Dominance

Mendelian

Gene another particular gene

Self-incompatibility

Mendelian

Gene another particular gene in a different species

Hybrid sterility

Mendelian

Gene the rest of the genotype

Dominance

Mendelian; statistical

Gene a specific environmental agent

Phenylketonuria

Mendelian; metabolic

Gene unspecified or multiple environmental factors

Eye facets;

Mendelian; statistical;

Colour pattern

Physiological

Genotype a specific environmental agent

Lung cancer

Statistical; GWA

Genotype unspecified or multiple environmental factors

Intelligence; height; fitness

Statistical; GWA

Genetic interaction was identified by Mendel (1866): dominance and recessiveness constitute interaction
between the two alleles of a gene.
Dominance and epistasis
Dominance, the phenomenon whereby one homozygote and the heterozygote of a diallelic gene are
indistinguishable phenotypically, was recognised by Mendel (1866) as an important attribute of the diploid
genotype, though not universal. His formulation made intelligible many phenomena observed but not
understood in animal and plant breeding (see e.g. Mayo 2009). It is true that dominance is a very general
Mendelian phenomenon, but it is purely somatic (Fisher 1918, p. 401). Falk (2001) has traced the history of
the use of dominance as an explanatory concept, and has concluded that it has little value apart from its
established descriptive value for phenotypes, where it is still very significant (e.g. Zaghloul et al. 2010).
Falks conclusion does not do justice to the role of dominance in quantitative genetics.
Fisher (1941) spelt out what was implicit in his discussion of adaptation in Fisher (1930): gene frequency
change under selection contains an essential G E interaction, since the average effect of a gene substitution
is dependent on gene frequency. Consider a diallelic locus at Hardy-Weinberg equilibrium:

Frequency
Genotypic value

A1A1

A1A2

A2A2

p2
a

2pq
d

q2
-a

The average effect of an allele, say A1, is the mean deviation from the population mean of individuals which
receive A1 from one parent and a random allele from the other parent. The average effect of A1 is given
by 1 = q(a+d(q-p)) and of A2, 2 = -p(a+d(q-p)). Then the average effect of a gene substitution is
= 1-2 = a+d(q-p). Unless A1A2 is exactly intermediate between the two homozygotes, is a function of
the gene frequency, and away from equilibrium always a function of the gene frequency (as Fisher originally
showed). The genotypic values can be changed to fitnesses without loss of generality (Fishers original
formulation), so that the environment in which gene frequency changes under selection includes the gene
frequency. Thus, the simplest genic interaction, that between alleles of one gene, is a G E interaction.
Fishers (1941) distinction between the average effect and the average excess of a gene substitution, whereby
the rate of change under selection depends on both, and hence on the association of a genes frequency with
the background genotype, provides that rationale for calling dominance a G E interaction. Much of this
was implicit also in Fisher (1918; see also the commentary by Moran and Smith 1966). Fisher (1918) wrote:
There is in dominance a certain latency. We may say that the somatic effects of identical genetic changes are not
additive, and for this reason the genetic similarity of relations is partly obscured in the statistical aggregate. A
similar deviation from the addition of superimposed effects may occur between different Mendelian factors. We
may use the term Epistacy to describe such deviation, which although potentially more complicated, has similar
statistical effects to dominance. If the two sexes are considered as Mendelian alternatives, the fact that other
Mendelian factors affect them to different extents may be regarded as an example of epistacy. (p. 404).
114

INTERACTION BETWEEN GENOTYPE AND ENVIRONMENT: A TALE OF TWO CONCEPTS

Fisher went on to show what components of variance could be measured to take account of additive effects,
dominance and epistacy, in the case of two segregating diallelic genes, and wrote that he had derived a
complete representation of any such deviations from linearity as may exist between two factors. Such dual
epistacy, as we may term it, is the only kind of which we shall treat. More complex interactions could
doubtless exist, but the number of unknowns introduced by dual epistacy alone, four, is more than can be
determined by existing data. In addition it is very improbable that any statistical effect, of a nature other than
that which we are considering, is actually produced by more complex somatic connections. (pp. 403-4).
Fisher (1930a, 1941) showed that directional natural selection would occur at a rate given by the additive
genetic variance, and it was the efficacy of natural selection that was his great concern. In writing earlier any
[other] statistical effect, Fisher was alluding to the fact that his analysis was conducted on the scale of the
original observations and their squares, so that higher order components were to be neglected. He returned
to this topic later, as will be discussed below.
First, however, we should return to the early history of genetical epistasis. Bateson (Bateson 1904, 1906,
1908, 1909a,b) considered that dominance and recessiveness were the result of the presence of the factor
(dominant) or its absence (recessiveness) and thought that human albinisms failure to follow Mendelian
segregation ratios might be the result of interactions, but also carried out or organised the first definitive
experiments that showed how more than one gene could influence a phenotype interactively. For example,
two pure-breeding lines of sweet peas with white flowers were crossed, producing an F1 generation with
purple flowers. When these were intercrossed, the F2 yielded purple:white::9:7. This could be explained as
an interaction between two diallelic genes, which was verified. The genotype cc of one gene with alleles C,
c yielded white no matter what the genotype of the second gene P, p, as did the genotype pp of the second
gene. Bateson called this interaction masking and introduced the term epistasis, which has been broadened
to cover different types of interaction, including others that Bateson investigated or suggested (e.g. a pair of
interacting genes leading to reproductive isolation and speciation: Bateson 1909b, Johnson 2009). (Comb
shape in poultry was the first case investigated and was also of this type.)
Interaction in a linear model
The previous section introduced a linear model for genotypic values in terms of the interactions between a
pair of alleles of one gene. Fisher (1918) considered additive effects, dominance and epistasis, together with
environment, as components of variance. Cockerham (1954), Kempthorne (1954) and others showed how the
epistatic variance could further be partitioned into AA, AD, DD etc. components. These developments
took the analysis further away from the underlying genetics and physiology, though identification of
dominance variance in the simpler partitioning can be useful (e. Rahman et al. 2009). Furthermore, the
algebraic elaborations obscured Fishers direct link to the variability on which natural selection would act in
any generation, though they became useful in plant breeding. They will not be considered further here.
In the context of the development of the design of experiments, Fisher and Mackenzie (1923) introduced
interaction generally in a linear model as follows:
yijk = + gi + cj + (gc)ij + ijk
where
yijk = yield of the ith potato variety receiving the jth fertiliser treatment in the kth plot receiving the
ijth variety-fertiliser combination
= overall mean yield
gi = effect of ith variety
cj = effect of jth fertiliser
(gc)ij = interaction between ith variety and jth fertiliser
ijk = plot effect.
115

O. MAYO

Fisher and also Yates generalised this model to the factorial design that allowed, in principle, detection of
any number of pairwise, threefold etc. interactions, at the expense of a rapid increase in the required size of
an experiment, hence fractional factorials and a whole new literature (see e.g. Bailey and Druilhet 2004). The
important point is that the word factorial was explicitly taken from Mendelian genetics, and the use of the
binomial expansion.
Hogbens contribution
Hogbens aim was to show that GE interactions were so great that they invalidated the linear/additive
approach to variance partitioning. Using others data, he showed that a major gene which influenced a
quantitative trait was unlikely to explain all the variance in one effectively constant environment, and that
there were major influences of environment on the genes effects. Unfortunately, in corresponding with
Fisher, he relied on an example in which, as Fisher correctly pointed out, the non-linearity of interaction
could be removed by a scale change. This weakened his argument in Fishers eyes. Indeed, Fisher hardly
bothered with Hogben again. (Questions of scale are still important in studies of GE interaction, of course
(e.g. Mani et al. 2008).)
Hogben also carried out the first analysis of correlations expected on the basis of X-linked inheritance, and showed
that these would rarely be important, as has proven to be the case (see e.g. Mayo, Brger and Leach 1990).
Hogben also recognised the importance of population stratification and how it could mimic assortation; this
is the basis of Guos (1999) model mentioned above. He considered that Fishers analysis of height dealt with
a sample of people from a relatively homogeneous UK subpopulation, insofar as a human group could be
homogeneous at that time, so that the conclusion could not apply to a population.
Hogben suggested within-family studies as a way of overcoming the problem of correlated environment
arising from systematic differences e.g. wealth. His results suggested genes were very important in
determining intelligence, so he rescued his hypothesis with an extra assumption ad hoc.
Overall, Hogben was correct in claiming that Fisher had gone beyond the data, but greatly underestimated
the utility and importance of Fishers linear modelling approach and novel statistical methods.
Furthermore, he recognised that in most biological systems, linearity would not be the rule, however well a
linear model might work as a first approximation. As noted above, it is unfortunate that in his discussions with
Fisher, he chose an example of non-additivity (non-linear effects of temperature and genotype on number of
eye-facets in mutant Drosophila melanogaster) that could be removed by a transformation of the data.
Application
As an example of a quantitative trait that might be simply described by a linear model, consider the amount
of wool an adult sheep grows each year, the clean scoured fleece weight, W. Turner (1958), considering how
the components of wool weight interacted to produce a fleece of a given weight, suggested that the sheep be
considered as a physical object with surface area A (proportional roughly to bodyweight B2/3 with an
allowance for wrinkled skin R). Wool fibres have an average diameter D, an average length measured by
staple length L with an allowance for crimping C and a certain specific gravity. There are N fibres per unit
area of smoothed skin. Fleece weight is then given by
W = k B2/3R2 LC4 (/4)D2 N,

(1)

where k includes specific gravity and a constant for unknown biological factors. Taking natural logarithms,
we expect
lnW = k + 2/3lnB + 2lnR + lnL + 4lnC + 2lnD + lnN

(2)

where k includes the other constants not shown. We can generalise with coefficients i where i = 1,2,3,4,5,6
for the six variables, expectations for four of the coefficients being as shown in the equation (2).
116

INTERACTION BETWEEN GENOTYPE AND ENVIRONMENT: A TALE OF TWO CONCEPTS

This formulation of Turners describes, in some sense, the average kinematics. When applied to the dynamics
of change in fleece weight, it cannot allow for the fact that fibres are derived from primary, secondary and
secondary derived follicles, each of which shows different correlations with the other components (e.g. Mayo
et al. 1994). If we split the follicle populations, we can no longer linearise the relationship as above; the
principle, however, is unchanged.
When we fit equation (2) by multiple linear regression, the coefficient of lnB can be significantly different
from 2/3, that of lnD significantly different from 2 etc., as shown in Table 2 for males and females from two
flocks selected differently for fleece weight (Mayo unpubl. 1964). These coefficients also differed from year
to year, on account of unknown interactions from seasonal effects such as nutrition, birth type (single, twin
etc.). The model cannot accommodate interactions, whether known environmental ones, or unknown genetical
ones, including pleiotropy (e.g. Mode and Robinson 1959).
Table 2. Estimates of partial linear regression coefficients for dependence of lnW (clean fleece weight) on its ln (components).
(See text for details of components and symbols).
Coefficient
Expectation

a1
1
(/)

a2
2
2

a3
3
1

a4
4
4

a5
5
2

a6
6
1

(Component

N)

Sex

Flock

M
M
F
F

I
V
I
V

No. animals
0.51
0.34
0.60
0.48

0.15
0.11
0.14
0.16

0.41
0.40
0.36
0.42

-0.18
-0.18
-0.09
-0.07

0.11
0.20
0.08
0.36

0.04
0.26
0.11
0.13

108
108
89
89

As discussed above, this physical model ignores the genetic architecture of fleece weight, some details of
which are shown in Table 3. It is noteworthy that fibre diameter D is the only trait whose causation appears
mainly to be genetical in this relatively benign environment. As noted by Fisher (see Bennett 1983, p. 260),
G E interaction in this kind of linear model is subsumed into the environmental component of variation.
If a sheep breeder wishes to select for increased wool production without loss of wool quality or other desirable
traits, the genetic architecture shown in Table 4 (extended to cover fertility etc.) can be used to construct a selection
index which will provide useful predictions for a small number of generations. However, correlated responses
are rarely close to prediction because the trait interrelationships are not linear (Franklin and Mayo 1998).
In the simpler situation of plant breeding trials, Yates and Cochran (1938) and others developed methods to
isolate the variety x season interactions, and as Freeman (1973) noted When interactions are present, estimates
of main effects are conditional: that is, one can only validly assert that genotypic effects are as observed in this
particular set of environments, not over all possible environments. (p. 346; Freemans italics). Thus, Hogbens
concerns are met in what is today standard practice in much experimental design, conduct and analysis.
Physiological interaction
Good examples for all of the categories in Table 1 exist. Colour in flowers forms a straightforward example
of gene x gene interaction as in the case described above. Diabetes forms an excellent example of the influence
of environment on disease incidence. Lung cancer and smoking constitute an archetypal field for G E studies.
Between the earliest demonstration that smoking habit was more concordant in MZ than DZ twins (e.g. Fisher
1958) and recent work elucidating this difference (e.g. Vink et al. 2009), not to mention work showing specific
interaction among genotype, smoking and autoimmunity to a specific enzyme (Mahdi et al., 2009), however,
lie multiple false trails. Table 5 illustrates the complexity of interactions associated with smoking habit.
117

O. MAYO
Table 3. Heritabilities (bold diagonal) and genetic (above diagonal) and environmental (below diagonal) correlations for W and
its components (Hancock et al. 1979; aGregory 1982a half-sib above dam-offspring estimate; bGregory 1982b).

0.12
0.33
0.25
0.12

0.08
-0.22
0.06
0.31

0.14b

0.17
0.54
0.61
0.80

-0.61
-0.58
-0.08
-0.37

0.06
0.04
0.23
0.08

0.17
0.53
0.05
-0.08

0.43
0.51
0.46
0.43

0.48
0.33
0.25
0.12

-0.26b

-0.12
0.20
0.07
0.40

0.07
-0.03
0.48
0.06

0.03
-0.23
-0.15
0.01

-0.34
-0.20
-0.41

0.22b

0.03b

0.29a
0.44a

-0.54b

0.20b

0.04b

0.15b

0.37
0.19
0.21
0.12

0.34
0.08
0.09
0.08

-0.16b

0.34
0.28
0.54
0.40

-0.08
-0.61
0.30
-0.55

-0.10
-0.12
0.48
-0.11

-0.17
0.08
-0.58
0.02

-0.24
-0.18
-0.33
-0.26

-0.05
0.01
-0.14
-0.02

0.08b

-0.31
-0.18
-0.52
-0.07

0.26
0.38
0.31
0.57

-0.23
-0.33
-0.51
-0.24

-0.11
-0.25
0.04
0.05

0.22

0.09

0.06b

0.42

-0.20

0.66

-0.44

0.26
0.13
0.00

0.13
0.22
0.00

0.28
-0.18
0.00

0.05
-0.10
0.00

0.66
0.51
1.03

-0.20
-0.63
-0.82

-0.03
0.23
0.15
0.32

0.08
-0.03
-0.09
0.05

-0.33
-0.10
-0.14
-0.19

0.05
0.04
0.02
-0.06

-0.34
-0.38
-0.31
0.00

0.33
0.31
0.16
0.46

0.02b

Table 4. Applications of genotype environment interaction in human health (modified from Thomas 2010).
Field

Application

Basic genetics

Elucidation of mechanisms
Gene identification

Pharmacogenetics/pharmacogenomics

Elucidating heterogeneity in response to drugs


Identifying subgroups for which drugs should/should not be
prescribed
Therapy through individual risk analysis

Ecogenetics

Elucidating heterogeneity in response to environmental agents


Basing public health measures on ecogenetic understanding
Disease prevention through individual risk analysis

118

INTERACTION BETWEEN GENOTYPE AND ENVIRONMENT: A TALE OF TWO CONCEPTS


Table 5. Odds ratio of having the adenosine A2A receptor (ADORA2A) 1083TT genotype for caffeine intake among non-smokers
and current smokers (Cornelis et al. 2007).
Caffeine intake
(mg/d)
All subjects
<100
100200
>200400
>400

ADORA2A genotype
CC
CT TT
(numbers of persons)

Odds ratio
(95% CI)

150
261
1062
426

100
129
446
161

1.00
0.74 (0.53, 1.03)
0.63 (0.48, 0.83)
0.57 (0.42, 0.77)

127
216
714
174

78
96
291
71

1.00
0.72 (0.50, 1.05)
0.66 (0.49, 0.91)
0.66 (0.45, 0.99)

23
45
348
252

22
33
155
90

1.00
0.77 (0.37, 1.66)
0.47 (0.25, 0.86)
0.37 (0.12, 0.70)

P for trend < 0.001


Non-smokers
<100
100-200
>200-400
>400
P for trend 0.03
Smokers
<100
100-200
>200-400
>400
P for trend < 0.001

Hogben was correct in his conclusions about genetical analyses of IQ and other similar human variables: they
are fraught with difficulty and have in the past frequently been invalid. I cannot find that Fisher carried any
out. Certainly he did not do so in 1918 or in his 1925 paper on twins (analysing others data) and 1928 paper
on triplets. Hogben (1933) did not note that in 1925 (in the paper he cited in the references but not in the
text) Fisher got a lower value from MZ twins and puzzled over it.
Hogben was also correct in saying that Fisher's 1918 conclusion, that variation in stature in Pearson's data was
totally explicable from genetical and assortative mating influences, probably reflected both the inclusion of
some G E effects and the relative homogeneity of environment of Pearson's sample. Hence Fisher's
conclusion was over-confident to the point of error, and this concern carries through to the biology underlying
much of Fishers (1930) social reasoning. (Fishers general conclusion, of course, stood: that man, like other
animals, owed his origin to an evolutionary process governed by natural law; that those mental and moral
qualities most peculiar to mankind were analogous, in their nature, to the mental and moral qualities of animals
and in their mode of inheritance, to the character of the human and animal body. p. 170).
Genotype environment in development
The development of an organism from fertilised ovum to larva, adult or otherwise differentiated stage
represents G E interaction in its clearest form: an organism that develops in one environment is normal,
that which develops in another is abnormal. Phenylketonuria is a very clear example: human individuals
lacking functional hepatic phenylalanine hydroxylase cannot develop normally unless they have a diet
deficient in phenylalanine. Given such a diet, they develop normally. The adult organism, however, meets
thousands of minor insults as it develops, and the adult reflects most of these insults, to a small extent. We
are generally not in a position to describe the interactions and their outcomes, and do not have robust models
equivalent to those of quantitative genetics to describe and predict the multidimensional outcomes of
interaction (Oates et al. 2009).
119

O. MAYO

Linkage disequilibrium
Association studies form one of the basic tools for the detection of genes having small effects on traits of
interest. G E is one of many factors than can influence both the detection of associations and the reality of
the associations detected. The phenomenon of hitch-hiking has long been known (Maynard Smith and
Haigh 1974), but recently it has been recognised that rare variants can cause the detection of associations
between a trait and a common polymorphism that do not represent any functional relationship at all (Dickson
et al. 2009). Experiments and surveys undertaken to detect and estimate G E and to relate it to understood
metabolic pathways etc. need to take this and many other factors into account (Thomas 2010).
Investigation of the genetics of human cognitive function today
Human intelligence, in all its manifestations, is strongly influenced by genotype. This is well established
(e.g. Plomin et al. 2001, Jacobs et al. 2007). While many genes are known with defective alleles that produce
severe intellectual deficits, none are known that have a large effect on normal intelligence. Those that have
been shown to have a consistent effect, e.g. COMT, the structural gene for catechol - O - methyl transferase
(Barnett et al. 2008), absorb a tiny fraction of the phenotypic variance in the measured trait. This is hardly
surprising, since strong selection in small populations leads rapidly to fixation (e.g. Brger 2000), and there
is evidence (from patterns of linkage disequilibrium in the human genome) for such recent selection in
humans, at a time when the human population was not large (e.g. Voight et al. 2006).
Plomin (Plomin et al. 2001, Butcher et al. 2008) carried out genome-wide assays of general cognitive ability
g and showed that no QTL absorbed as much as 0.4% of the variance in g. They wrote in the second paper:
It is likely that QTL effect sizes, even for highly heritable traits such as cognitive abilities and disabilities,
are much smaller than previously assumed. Other conclusions are possible.
Data from experimental organisms reveal that classical Fisherian traits, i.e. those determined by very large
numbers of genes of small individual effect (though see Fisher 1940 for a better understanding of Fishers
views), are not at all rare (e.g. Mayo 2004). Interactions between QTL are more frequent than the QTL
themselves. How we use these findings is another matter (Table 2). For a general review of the problems
inherent in these studies, see Ioannidis (2007).
It is noteworthy that mechanisms may exist at the neural level to generate social behaviours directly (Tricomi
et al. 2010), such mechanisms being likely to influence apparent and even assessed intelligence. Investigation
of the physiological basis of cognitive function is accelerating (e.g. Deary et al. 2010), but the nuances of G
E will not yield readily, for example the recent media reports of differences in how the brain processes
language in native Mandarin and English speakers (and see also Varki et al. 2008).
Concluding note
Hogbens recognition of non-linearity has not prevented the widespread use of linear methods because they
work well to a first approximation. However, the volume of data available following the Human Genome
Project and the computing power available jointly mean that many classical problems must be revisited. In
1930, Fisher proved the important fundamental theorem of natural selection: The rate of increase in fitness
of any organism at any time is equal to its [additive] genetic variance [VA]. See Ewens (1989) for an
assessment of the importance of the theorem.
In 1930, nothing was known of VA in natural populations, though of course Fisher (e.g. 1930b) was interested
in obtaining evidence. Today, we almost always find VA > O for fitness traits; the question why VA has not
been exhausted by natural selection has frequently been asked, though it does not always require an answer
(e.g. Mayo et al., 1990). It may in future be answered generally by approaches such as that of Hine et al.
(2009). They use tensor geometry, first introduced into population genetics in 1940 by Claude Shannon, the
120

INTERACTION BETWEEN GENOTYPE AND ENVIRONMENT: A TALE OF TWO CONCEPTS

father of communication theory, though his work was barely noticed by the population genetics community.
On the point in question, Hine et al. (2009) wrote it is hard to escape the conclusion that we have yet to
come to an empirical understanding of the most basic aspects of how selection affects genetic variance in
natural populations.
Acknowledgements
I thank Ian Franklin, George Fraser, Carolyn Leach, Nick Martin and John Sved for helpful comments.
References
Bailey, R. A. and Druilhet, P. (2004) Optimality of neighbor-balanced designs for total effects. Annals of Statistics 32: 16501661.
Barnett, J. H. Scoriels, L. and Munafo, M. R. (2008) Meta-analysis of the cognitive effects of the catechol-O-methyltransferase
polymorphism. Biological Psychiatry 64: 137144.
Bateson, W. (1904) Experimental studies in the physiology of heredity. Reports to the Evolution Committee of the Royal Society 2:
1154.
Bateson, W. (1906) Experimental studies in the physiology of heredity. Reports to the Evolution Committee of the Royal Society 3:
153.
Bateson, W. 1908 Experimental studies in the physiology of heredity. Reports to the Evolution Committee of the Royal Society 4:160.
Bateson, W. (1909a) Mendel's Principles of Heredity. Cambridge University Press.
Bateson W (1909b) [Contribution to discussion of] The influence of heredity on disease, with special reference to tuberculosis,
cancer and diseases of the nervous system. Proceedings of the Royal Society of Medicine 81: 225229.
Bateson W (1909c) Heredity and variation in modern lights. Pp. 85101 In Darwin and Modern Science (ed. AC Seward) Cambridge
University Press.
Bennett, J. H. (editor) (1983). Natural selection, heredity and Eugenics including selected correspondence of R. A. Fisher with
Leonard Darwin and others. Oxford University Press.
Butcher, L. M., Davis, O. S. P., Craig, I. W. and Plomin, R. (2008) Genome-wide quantitative trait locus association scan of general
cognitive ability using pooled DNA and 500K single nucleotide polymorphism microarrays. Genes Brain Behavior, 2008 7: 435
446.
Cockerham, C. C. (1954) An extension of the concept of partitioning the hereditary variance for analysis of covariances among
relatives when epistasis is present. Genetics 39: 859882.
Cornelis, M. C., El-Sohemy, A. and Campos, H. (2007) Genetic polymorphism of the adenosine A2A receptor is associated with
habitual caffeine consumption. American Journal of Clinical Nutrition 86: 2404.
Deary, I. J., Penke, L. and Johnson, W. (2010) The neuroscience of human intelligence differences. Nature Reviews Neuroscience
11: 201211.
Dickson, S. P., Wang, K., Krantz, I., Hakonarson, H. and Goldstein D. B. (2009) Rare variants create synthetic genome-wide
associations. PLoS Biology 8: e1000294 (112).
Ewens, W. J. (1989) An interpretation and proof of the fundamental theorem of natural selection. Theoretical Population Biology 36:
6780.
Falk, R. (2001) The rise and fall of dominance Biology and Philosophy 16: 285323.
Fisher, R. A. (1918) On the correlation between relatives on the supposition of Mendelian inheritance. Transactions of the Royal
Society of Edinburgh 52: 399433.
Fisher, R. A. (1930a) The Genetical Theory of Natural Selection. Oxford University Press.

121

O. MAYO
Fisher, R. A. (1930b) Mortality among plants and its bearing on natural selection. Nature 125: 972973.
Fisher R A (1941) Average excess and average effect of a gene. Annals of Eugenics 11: 5363.
Fisher, R. A. (1958) Lung cancer and cigarettes? Nature 182: 108.
Franklin, I. R. and Mayo, O. (1998) The place of QTL in the basis of quantitative genetics. II. Mapping QTL from a wide cross. In
Proceedings of the 6th World Congress on Genetics Applied to Livestock Production. Armidale, NSW, Australia: Organising
Committee, 6th World Congress on Genetics Applied to Livestock Production, 1998: pp. 273276.
Freeman, G. H. (1973) Statistical methods for the analysis of genotype-environment interactions. Heredity 31: 339354.
Galton, F. (1883) Inquiries into Human Faculty and its Development. London, J. M. Dent (p. 128 of 1908 reprint).
Gregory, I. P. (1982a) Genetic studies of South Australian Merino sheep. IV. Heritabilities of various wool and body traits. Australian
Journal of Agricultural Research 33: 355362.
Gregory, I. P. (1982b) Genetic studies of South Australian Merino sheep. IV. Genetic, phenotypic and environmental correlations
between various wool and body traits. Australian Journal of Agricultural Research 33: 363373.
Guo, S.-W. (1999) The behaviors of some heritability estimators in the complete absence of genetic factors. Human Heredity 49:
215228.
Hancock, T.W., Mayo, O. and Brady, R.E. (1979) Response to partial selection on clean fleece weight in South Australian strong-wool
Merino sheep IV. Heritabilities and genetic correlations for fleece weight and its components. Australian Journal of
Agricultural Research 30: 17389.
Hine, E., Chenoweth, S. F., Rundle, H. D. and Blows, M. W. (2009) Characterizing the evolution of genetic variance using genetic
covariance tensors. Philosophical Transactions of the Royal Society of London B 364: 15671578.
Hogben, L. (1933) The limits of applicability of correlation technique in human genetics. Journal of Genetics 27: 379406.
Ioannidis, J. P. A. (2007) Non-replication and inconsistency in the genome-wide association setting. Human Heredity 64: 203213.
Jacobs, N., van Os, J., Derom, C. and Thiery, E. (2007) Heritability of intelligence. Twin Research and Human Genetics 10: 1114.
Johnson, N. A. (2009) One hundred years after Bateson: A pair of incompatible genes underlying hybrid sterility between species.
Heredity 103: 360360.
Kempthorne, O. (1954) The correlation between relatives in a random mating population. Proceedings of the Royal Society Series
B 143: 103113.
Mahdi, H., Fisher, B. A., Kllberg, H., Plant, D., Malmstrm, V., Rnnelid, J., Charles, P., Ding, B., Alfredsson, L., Padyukov, L.,
Symmons, D. P. M., Venables, P. J., Klareskog, L. & Lundberg, K. (2009) Specific interaction between genotype, smoking and
autoimmunity to citrullinated -enolase in the etiology of rheumatoid arthritis. Nature Genetics 41: 13191326.
Mani, R, St Onge R P, Hartman J L, Giaever, G. and Roth F P (2008) Defining genetic interaction. Proceedings of the National
Academy of Science 105: 34613466.
Maynard Smith, J., and Haigh, J. (1974). The hitch-hiking effect of a favourable gene. Genetical Research 23: 2335.
Mayo O. (1964) Unpublished B. Sc. (Hons) thesis, University of Adelaide.
Mayo, O., Brger, R. and Leach, C.R. (1990). The heritability of fitness: Some single gene models. Theoretical and Applied Genetics
79: 278284.
Mayo, O., Crook, B.J., Lax, J. Swan, A. and Hancock, T.W. (1994). The determination of fibre diameter distribution. Wool Technology
and Sheep Breeding 42: 231236.
Mayo, O. (2004) Interaction and quantitative trait loci. Australian Journal of Experimental Agriculture 44: 11351140.
Mayo, O. (2009) The influence of animal breeding on The Origin of Species by Charles Darwin. Proceedings of the Society for the
Advancement of Animal Breeding and Genetics 18: 711.

122

INTERACTION BETWEEN GENOTYPE AND ENVIRONMENT: A TALE OF TWO CONCEPTS


Mendel G. (1866) Versuche ber Pflanzen-Hybriden. Verh. Naturforsch. Ver. Brnn 4: 347.
Mode, C. J. and Robinson, H. F. (1959) Pleiotropism and the genetic variance and covariance. Biometrics 15: 518537.
Moran, P.A.P. and Smith, C.A.B. (1966) Commentary on R. A. Fishers paper on the correlation between relatives on the supposition
of Mendelian inheritance. Eugenics Laboratory Memoirs 61. Cambridge University Press.
Oates, A. C., Gorfinkiel, N., Gonzlez-Gaitn, M. and Heisenberg, C.-P. (2009) Quantitative approaches in developmental biology.
Nature Reviews Genetics 10: 517530.
Plomin, R., Hill L., Craig I.W., McGuffin P., Purcell S., Sham P., Lubinski D., Thompson L.A., Fisher, P.J., Turic, D. and Owen,
M.J. (2001) A genome-wide scan of 1842 DNA markers for allelic associations with general cognitive ability: A five-stage
design using DNA pooling and extreme selected groups. Behavior Genetics 31: 497509.
Rahman, I., Bennet, A. M., Pedersen, N. L., de Faire, U., Svensson, P. and Magnusson, P.K.E. (2009) Genetic dominance influences
blood biomarker levels in a sample of 12,000 Swedish elderly twins. Twin Research and Human Genetics 12: 286294.
Shannon, C. E. (1940) An algebra for theoretical genetics. Unpublished Ph.D. thesis, Massachusetts Institute of Technology.
http://dspace.mit.edu/bitstream/handle/1721.1/11174/34541447.pdf?sequence=1 (accessed 10 March 2011)
Tabery, J. 2008 R. A. Fisher, Lancelot Hogben, and the origin(s) of genotype-environment interaction. Journal of the History of
Biology 41: 717761.
Thomas, D. 2010 Gene-environment-wide association studies: Emerging approaches. Nature Reviews Genetics 11: 259272.
Tricomi, E., Rangel, A., Camerer, C. F. and ODoherty, J. P. 2010 Neural evidence for inequality-averse social preferences. Nature
463: 10891092.
Varki, A., Geschwind, D.H. and Eichler, E.E. (2008) Explaining human uniqueness: genome interactions with environment, behaviour
and culture. Nature Reviews Genetics 9: 749763.
Vink, J.M., Smit, A.B., de Geus, E.J., Sullivan, P., Willemsen, G., Hottenga, J.J., Smit, J.H., Hoogendijk, W.J., Zitman, F.G., Peltonen,
L., Kaprio, J., Pedersen, N.L., Magnusson, P.K., Spector, T.D., Kyvik, K.O., Morley, K.I., Heath, A.C., Martin, N.G.,
Westendorp, R.G., Slagboom, P.E., Tiemeier, H., Hofman, A., Uitterlinden, A.G., Aulchenko, Y.S., Amin, N., van Duijn, C.,
Penninx, B.W. and Boomsma, D.I. (2009) Genome-wide association study of smoking initiation and current smoking. American
Journal of Human Genetics 84: 367379.
Voight, B.F., Kudaravalli, S., Wen, X. and Pritchard, J.K. (2006) A map of recent positive selection in the human genome. PLoS Biol
4(3): e72.
Yates, F. and Cochran, W. G. (1938) The analysis of groups of experiments. Journal of Agricultural Science 28: 556580.
Zaghloul, N. A., Liu, Y., Gerdes, J. M., Gascue, C., Oh, E. C., Leitch, C. C., Bromberg, Y., Binkley, J., Leibel, R. L., Badano, J. L.
and Katsanis, N. (2010) Functional analyses of variants reveal a significant role for dominant negative and common alleles in
oligogenic Bardet-Biedl syndrome. PNAS 107: 106027.

123

Das könnte Ihnen auch gefallen