Opinion

TRENDS in Plant Science

Vol.12 No.10

Parental conflict does not necessarily

lead to the evolution of imprinting
Tom J. de Jong1 and Rod J. Scott2
1
2

Institute of Biology, Leiden University, Leiden, PO Box 9516, 2300 RA Leiden, The Netherlands
University of Bath, Department of Biology and Biochemistry, Bath, BA2 7AY, UK

For some genes, the epigenetic state (whether they are

expressed) depends on whether the gene is inherited
through the mother or the father. Such imprinting, or
parent-specific gene expression (PSGE), occurs in mammals, including humans, and higher plants. The theory
that PSGE solves genetic conflict between mother and
father is widely accepted. We argue, however, that the
conditions for PSGE to evolve are restricted. With
respect to seed size, PSGE can only evolve when the
developing offspring has a strong effect on its own
resource acquisition. When seed size is close to the
optimum for the maternal parent, there is no internal
conflict in the offspring because maternally and paternally derived genes both favour increased seed size.
Although the literature generally suggests that the
maternal parent controls seed size, a number of observations suggest an additional role for the paternal
parent. Here, we critically evaluate these studies and
suggest a rigorous methodology for establishing
paternal effects on seed size, which can be applied to
the model species Arabidopsis thaliana.
Conflicts over seed size
In the simplest optimisation models, it is assumed that
genes expressed in the maternal parent control the size of
all seeds [1,2]. When genes expressed in the developing
offspring can affect their own resource acquisition, alleles
that acquire more resources pass on more copies of themselves than do neutral alleles. This shifts seed size to
higher values than would be desirable for the mother
(parentoffspring conflict [3]). The less related competing
offspring are to one another, the larger the advantage to an
allele in taking more resources for itself and the larger the
desired seed size for the offspring.
The theory of parentoffspring conflict (see Glossary) is
complemented by parental conflict or kinship theory (see
Glossary) [4]. This theory assigns different roles to maternally and paternally derived alleles in the offspring. Given
that relatedness is higher for the maternally than for the
paternally inherited allele (Box 1), a conditional strategy
seems the best solution: take many resources when paternally inherited, but few when maternally inherited. This
can be achieved by parent-specific gene expression (PSGE;
see Glossary), that is, by switching alleles on or off, depending on the parent of origin. Thus if degrees of relatedness
Corresponding author: de Jong, T.J. (t.j.de.jong@biology.leidenuniv.nl).
Available online 12 September 2007.
www.sciencedirect.com

differ for the maternal and paternal genomes as they

often will PSGE is expected to evolve [5].
We will argue that the conditions for PSGE to evolve are
more restricted. What is needed is a considerable effect of
the offspring on its own size. We will then review how seed
size depends on genotype of the maternal parent and
genotype of the offspring. Because effects of pollen parent
on seed size are subtle and can easily be obscured by other
factors, we will end by suggesting a methodology that
avoids several caveats.
A more precise model
The quotation cited above [5] suggests that conditions for
the evolution of PSGE are easily met. Nevertheless, a
problem becomes apparent when the argument is quantified (Box 1). On a fully selfing plant, the relatedness (r)
between seed embryos equals one, and the desired seed
sizes for parent (point A) and offspring coincide. With
certain outcrossing, r is smaller than one and the optimal
seed size under offspring control (C) exceeds that under
maternal control. We can also calculate relatedness for the
maternally inherited allele in the embryo and the corresponding seed size B. The same holds for the paternally
inherited allele and the corresponding seed size D. We now
have four desired offspring sizes (Figure 1), depending on
who is in control. The desired offspring sizes rank
A<B<C<D (Box 1). When seed size lies between A and
B, PSGE is not selected for. An allele for slightly bigger
Glossary
Coefficient of relatedness (r): for outbreeding diploids rJK is the probability that
an allele in individual J is also present by identical descent in individual K. Two
full sibs have a relatedness of 0.5. For inbreeding populations, the computation
becomes more complex [28].
Parent-offspring conflict theory: selection for a certain character, such as seed
size, differs depending on whether the gene for this character is expressed in
the parent or in the developing offspring. The offspring is selected to take more
than the parent is willing to give, and the theory allows a prediction of the
desired optima, based on the principle of natural selection.
Parental-conflict theory or kinship theory: unless there is 100% selfing, the
paternally derived allele in the embryo is less related to other embryos on
the same plant as the maternally derived allele and is therefore selected to
acquire more resource at the expense of other seeds. For instance, consider an
outbreeding population and two fathers siring each 50% of the seeds on one
maternal parent. For two randomly chosen seeds, the average relatedness is
0.50 for the maternal allele and 0.25 for the paternal allele. Relatedness (r)
between two seed embryos is 0.375. Using these coefficients of relatedness,
the theory predicts the different optima. Imprinting can solve the problem of
desiring more as the paternal allele, than as the maternal allele in the offspring.
Imprinting or parent-specific gene expression (PSGE): switching on or off
genes for resource acquisition depending on whether the gene is maternally or
paternally inherited.

1360-1385/$ see front matter 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.tplants.2007.07.003

Opinion

440

TRENDS in Plant Science

Vol.12 No.10

Box 1. Calculating optimal seed size

We begin with the relation between offspring size (x) and fitness per
offspring [f(x)] (Figure I). Big seeds often have better survival, but the
shape of the curve is arbitrary [28]. The optimal (desired) seed size for
the mother is found by drawing a line through the origin that touches
the survival graph [1] at point A where the slope of the line equals f(x)/x.
At this point, fitness returns per unit of investment are maximal. With
offspring control the desired size is C, where the slope equals rf(x)/x
[28], where r is the coefficient of relatedness (see Glossary) between
competing embryos.
Relatedness can be calculated separately for the maternal (m) and
paternal ( p) allele in the offspring. When only the maternal allele
matters for offspring size, the desired size is B where the slope equals
mf(x)/x. When only the paternal allele matters, the desired size is D,
where the slope equals pf(x)/x. For two embryos on a mother plant
with selfing rate S and n competing fathers, relatedness can be
computed using these Equations IIII [28]:
m

0:251 S 4  S  S 2
2S

[Equation I]

0:251 S 3S  S 2 1  S2

n2  S
2  S

[Equation II]

r m p=2

[Equation III]

Without selfing (S=0) this reduces to m=0.5 and p=0.5/n [29]. Unless
there is full selfing, it can be shown that m>p, a logical result that
underlies parental conflict theory. The maternal allele in the offspring
is never selected to acquire more resource than is the paternal allele,
and the four desired offspring sizes rank as A<B<C<D. Two further
predictions are less obvious.
First, a threshold (the difference between A and B) needs to be
crossed before selection favours PSGE. The threshold is small and
unimportant if m is close to one and steep if m reaches its lowest
value (0.5) with full outcrossing.
Second, the intensity of imprinting [30] increases with the
difference between m and p. This difference increases with the
number of competing fathers but also depends on the selfing rate. By
differentiating (m-p) with respect to S, it can be shown that the largest
difference between m and p with one father (n=1) is found at S=0.268
and with many fathers at S=0.586.
These two theoretical predictions go against some common
(mis)interpretations of parental conflict theory. The threshold neces-

seeds should be expressed in both the maternally and

paternally derived allele in the offspring. When seed size
lies between B and D, PSGE is selected for. Conditions for
PSGE are thus restricted, contrary to common belief [5]. It
would be possible to make the model more complicated by
including genes of large effect and examine when PSGE of

Figure 1. Four desired seed sizes, corresponding to full control by either genes
expressed in the mother plant (A), only the maternal allele in the offspring (B), both
alleles in the offspring (C), and only the paternal allele in the offspring (D). PSGE is
selectively favourable when actual seed size is between points B and D. When seed
size is below threshold B, the maternal and paternal alleles in the offspring are
both selected to acquire more resources.
www.sciencedirect.com

sitates that conditions for PSGE to evolve are restricted. Furthermore,

the expectation that PSGE can only evolve in fully outcrossing species
[17,23,30] and not in species with an intermediate selfing rate is
unfounded. The view that PSGE in Arabidopsis thaliana is not
selected for and is perhaps a remnant from an outcrossing ancestor
[23] is also not backed by theory. A. thaliana is 100% selfing when
plants are left to themselves in the laboratory. However, under
favourable weather conditions in the field small insects actively visit
the flowers and can induce outcrossing [31]. With, for instance, a
selfing rate of 90%, m is close to one (Equation I), consequently
threshold B is low and even a small effect of the offspring on its own
size will induce conditions under which PSGE is favoured by
selection.

Figure I. Survival as a function of seed size x with full outcrossing (S=0) and
many competing fathers (r=0.25; m=0.5; p=0). Point A gives the optimal
offspring size under maternal control, where: df(x)/dx = f(x)/x. Point C gives
the desired size for the offspring, where: df(x)/dx = rf(x)/x, in which r is the
average relatedness between embryos from the same mother [28]. Similarly
points B and D correspond to desired seed sizes for the maternally and
paternally inherited allele in the offspring, respectively. Note that in this
example the offspring needs to pull seed size from 3.45 to more then 8.58 to
induce conditions favourable for the evolution of PSGE. Note also that in this
example D is very large. Figure reproduced, with permission, from Ref.
[28].

these genes is favoured by natural selection. But this should

not distract from the basic reasoning presented here. The
important biological question that emerges is: can the offspring enhance seed size to such an extent that threshold B
is crossed and PSGE can evolve? Only a clear yes to this
question is in line with the kinship theory for the evolution of
PSGE. No or maybe means that we cannot claim that
parental conflict explains imprinting and will need to study
in more detail how conflicts over seed size are resolved.
Effects of offspring genotype on seed size
The most direct way to estimate offspring effects on seed
size is to highlight the role of the father. Unlike maternal
parents that can affect seed size in several ways, paternal
parents can only influence seed set through effects in the
endosperm and possibly the embryo. Hence, the most
compelling evidence for offspring control over seed size
would be obtained if the fertilization by different fathers
resulted in differences in seed size. Relevant studies occur
in different fields.
Ecological studies
Some ecological studies partitioned the variance in seed
mass into different components. An extensive study on

Opinion

TRENDS in Plant Science

baby blue eyes (Nemophila menziesii) involved crossing 16

parents and screening more than 11 000 progeny over
three generations [6]. The maternal genotype accounted
for 20% of the variation in seed mass, showing that there is
an appreciable potential for selection on seed mass through
a maternal effect. The choice of the paternal parent had no
effect on seed size. In the partridge pea, Chamaecrista
fasciculata, differences in seed size existed between selfing
and outcrossing treatments [7]. However, within the outcrossing treatment, no detectable variation among male
pollen donors was found in their ability to accrue resources
from the seed parent.
Crosses in Arabidopsis thaliana
In reciprocal crosses between the small seeded Landsberg
erecta (Ler) genotype of A. thaliana and the large-seeded
genotype from the Cape Verde Islands (Cvi), seed size
resembled that of the mother [8]. Some effect of the father
appeared to be present in the crosses but this effect had no
clear direction. A similar result was observed in reciprocal
crosses between Col-3 and the large seeded mutant megaintegumenta (mnt) [9]. A recent study [10] used 13 accessions to sire seeds on female (male sterile) Ler plants. The
length of the hybrid seeds was then compared with the
length of seeds on the paternal parents when they were left
to self-pollinate. Hybrid seeds produced on female plants
were longer than selfed seeds on hermaphrodite plants,
which led the authors to conclude that this reflected heterosis. This conclusion is, however, doubtful because selfed
seeds on female Ler were also bigger (16%) than selfed
seeds on hermaphrodite Ler, and this observation could
have easily explained the above result. Finally, fertilisation with pollen from another large-seeded species also did
not increase seed size in A. thaliana (Figure 2).
Xenia
In the agricultural and horticultural sciences, the effect of
the pollen parent on characters of the seed or fruit is known

Figure 2. Cross between Arabidopsis thaliana (mother) and the outcrossing

Arabidopsis lyrata subsp. petraea (father). The average seed weight in
micrograms is provided for each parent and the offspring (F1) (R.J. Scott,
unpublished data).
www.sciencedirect.com

Vol.12 No.10

441

as xenia [11]. Some effects of the pollen parent are quite

spectacular. For instance, in mandarin Citrus reticulata,
certain pollen parents affect the sugar content of the fruit
[12], which is composed of maternal tissue. In general the
xenia literature suggests no clear effects of pollen parent
on seed size. Two exceptions to this rule need to be mentioned. Several authors found paternal effects on kernel
weight in corn Zea mays, which are in the order of 3.9% [13]
or 4.112.6% [14]. A vintage study on different varieties of
garden bean Phaseolus vulgaris concluded that, in most
crosses, hybrid seeds appeared identical to seeds on the
mother plant [15]. However, in crosses between Marblehead (average seed mass 0.694 g) and Powell Prolific
(0.420 g), effects of the paternal parent were observed.
With Powell Prolific as mother, seeds were of standard
weight (0.436 g), whereas with Marblehead as mother and
Powell Prolific as father, seeds became larger (1.076 g). All
hybrids grew vigorously, producing more pods than their
parents. One could think that the bigger seeds are also due
to heterosis. However, the xenia effect only occurred in one
crossing direction, so that it cannot simply be a result of
increased heterozygosity in endosperm or embryo.
Crosses between plants with different ploidy levels
With few exceptions [1315] the previous examples
suggested only a small effect of pollen parent on seed mass.
Conflicting data come from interploidy crosses. Excess of
paternal genomes resulted in more vigorous endosperm
development [16] and, provided seeds are not aborted, in
bigger seeds [17].
In A. thaliana mean seed weight on a diploid mother
increased from ca. 21 mg for a diploid father to ca. 53 mg for
a tetraploid father [17]. Excess of maternal genomes has
the opposite effect. Seeds of a tetraploid mother fertilised
by a diploid father weighed only 13.5 mg [17]. These results
show that, through this major change in endosperm genotype, the offspring can affect its own size. Here, the mother
appears to have lost control by allowing both the formation
of very large seeds and, in the reciprocal direction, small
seeds with low chances of survival.
Arabidopsis mutants
Heterozygous medea (mea) mutants [18] of A. thaliana
produce 50% seeds with overproliferation of the endosperm, as observed in crosses with polyploid fathers.
Despite their initial vigorous development, these seeds
are all inviable, whereas the other 50% develop normally.
When mea mutants are used as pollen parents, all seeds
develop normally, showing that the mutant allele has no
effect when supplied in the pollen [18]. The wild-type MEA
allele inhibits endosperm development. It is silent from the
paternal allele in endosperm, but active from the maternal
allele owing to the methylation-removing activity of
DEMETER [19]. Although mea is not a gene for seed size,
given that eventually all mutant seeds die, the results
show that a mutation at a regulatory gene can have a
strong impact on endosperm proliferation and, perhaps,
hypomorphic mea alleles or naturally occurring MEA
alleles lead to less imprinting and an effect on seed size.
Several research groups [20,21] used mutants
with hypomethylated DNA in crossing studies. When

442

Opinion

TRENDS in Plant Science

hypomethylation increases the number of genes expressed,

different effects would be expected depending on the
parent of origin. When inhibitors are off in the paternal
genome in the offspring, switching on these genes reduces
seed size. When promoters are off in the maternal genome,
switching on these genes increases seed size. Indeed, in
A. thaliana, the METHYLTRANSFERASE I antisense
transgene (METI a/s) produced smaller seeds (9 mg) then
normal (21 mg) when paternally transmitted and larger
seeds (32 mg) when maternally transmitted [20]. Parent-oforigin effects obtained with other hypomethylated mutants
[21] were in the same direction.
Few imprinted genes have so far been discovered in
plants [22]. MEA and other imprinted genes in the FIS
complex are maternally expressed [23]. Only a single,
paternal enhancer of endosperm growth, PHERES1, has
so far been found in A. thaliana and this gene might be an
antagonist of MEA [24].
Whether any (diploid) father sires significantly bigger
seeds than those that result from selfing, is something that
still needs to be shown convincingly. The results from
A. thaliana are generally considered to be consistent with
parental conflict theory, which is correct provided that seed
size lies between point B and D in Figure 1, that is, the
offspring exerts some effect on its own size.
Methodological problems in estimating seed size
There are more examples, especially in the xenia
literature, in which pollen-parent effects were reported.
However, it is not the purpose here to list these studies

Vol.12 No.10

extensively; often the methods for obtaining seeds are not

explained in sufficient detail or other effects confound the
effect of the paternal parent.
For instance, seed sizes after self-fertilization were
often obtained from plants that were simply allowed to
self-pollinate. It is problematic to compare these seed sizes
with those obtained in a single fruit after enforced outcrossing. When outcrossing involves emasculation of the
flower before hand pollination, these effects become confounded with the effects of the pollen parent. When outcrossing involves pollinating a subset of the flowers on
male-sterile plants, such individuals are likely to have
more resources available and to produce heavier seeds,
than hermaphrodite plants with self-pollination in all
flowers [10].
In many species, A. thaliana included [8,9], a trade-off
exists between the number of seeds produced and the
weight per seed. When one father produces less pollen
or pollen of inferior quality than other fathers, the size
number trade-off leads to the first father siring fewer but
bigger seeds. Such results do not reflect genetic differences
between paternal parents in the ability to acquire
resources. In bladder campion Silene vulgaris, the first
seeds to be formed are bigger because they are first to tap
into maternal resources. Experimental manipulation
showed that the order of fertilisation matters here and
not so much differences in genetic quality between offspring [25]. To minimise error when estimating seed size
we advocate standardizing methods, and Box 2 provides
some suggestions for A. thaliana.

Box 2. Protocol for detecting pollen-parent effects on seed size in Arabidopsis thaliana
Because the seed parent has a potentially large influence over seed
size, successful detection of a pollen-parent-mediated effect and
comparison of effects of different fathers, demands careful control
over several factors. The most potent variables are seed number per
pod (reducing seed number increases seed weight; Figure I) and pod
number per plant (reducing pod number per plant also increases
seed weight). Therefore, ensuring comparable seed number per pod
and pod number per plant is important.
The following experimental design termed restricted pollination
has proved effective [9]. Male-sterile seed parents selected from
either Col-0, Ler or C24 have been modified to carry a tapetumspecific barnase gene that provides excellent male sterility [32]. Each
pollen parent is tested by pollinating at least three flowers on the
main inflorescence of the seed parent. Fertilisation with pollen from
the hermaphrodite parent of the same accession can serve as a
control and a reference point for comparison between different
mother plants. Because early flowers are often of poor quality, the
first flower selected is generally the tenth from the base of the
inflorescence. Mature flowers have fully reflexed petals and stigmas
with long papillae. The remainder of the inflorescence is removed
with watchmakers forceps before pollinating the flowers. Pollen is
applied gently because crushing the stigmatic papillae dramatically
reduces seed set. Sufficient pollen to sire at least 50 seeds must be
applied, which usually requires pollen from one flower per stigma.
Secondary inflorescences are continuously removed to ensure all
seed parents are kept as similar as possible. At collection, mature
pods have a yellow to purple colour, and split easily when gently
squeezed.
Where a male-sterile seed parent is not available, flowers are
manually emasculated. The largest closed buds, centred around
the tenth flower on the main inflorescence are selected. The rest of the
inflorescence is removed, together with all the flowers and pods
below the target buds. The buds are gently opened using forceps to
www.sciencedirect.com

part the sepals and petals (no need to remove these organs), and the
six stamens removed by gently closing the forceps on the stamen
filament and lifting out the anther. Emasculated flowers are matured
for 3648 h before applying pollen. It is essential to check for pollen
already on the stigma, and that the papillae are elongated but not
withered.

Figure I. Average seed mass decreases with seeds per fruit in crosses between
tetraploid A. thaliana (C24). The correlation is significant (r=0.44; P<0.001) and
when it is fitted by linear regression the slope is 0.216, that is, for each extra
seed per fruit, the average seed size decreases with 0.216 mg (R.J. Scott,
unpublished data). For a similar relation for diploids, see Ref. [8].

Opinion

TRENDS in Plant Science

Table 1. Effect of maternal and paternal parent on nut size in

almond, Prunus dulcis, cultivars as compared to open
pollinationa,b
Maternal parentc,d
Small
Medium
Large

Paternal parent:
Small
Medium
4.03
4.38
5.24
5.15
8.35
8.45

Large
4.37
5.30
8.59

Nut size: length X narrowness (cm).

See Ref. [26].
Cultivars: small, Drake; medium, Merced; large, Ne Plus Ultra.
d
Open pollination: gray shaded figures.
b
c

Other caveats are of a statistical nature. Seeds within a

fruit are not independent of each other. For the statistical
analysis this requires keeping separate seeds from different fruits and different maternal parents, even if these
parents are genetically the same. Ideally the crossing
design is balanced, using all combinations of genotypes.
If we take only the three cultivars that were repeatedly
used in crossing experiments with almond [26] the conclusion must be that the genotype of the mother matters
most for seed set (Table 1). Yet the study from which
Table 1 was drawn emphasized a xenia effect on seed size,
based on additional fathers used in only few of the crosses
[26].
Concluding remarks
Does the genotype of the offspring matter for its own size; for
now the answer to this question is maybe. Therefore we
advocate paying more attention to the role of the father in
studies of seed size. Lack of variation in the ability of the
pollen parent to affect seed size does not mean that the
father is unimportant. Paternally derived genes are likely to
be under strong selection to acquire as much resource as
possible, eliminating all genetic variation in this ability from
the population. There could still be variation between isolated populations (or species) in paternal effects on seed size
and careful screening of many fathers from outside the
population might reveal this variation. Parental conflict
theory only predicts PSGE when the offspring pulled seed
size past threshold B (Box 1). Because this threshold B is low
in the mostly selfing A. thaliana, imprinting can evolve in
this species, even if the offspring has only a small effect on its
own size. However, in other species, specifically outcrossers
with necessarily a high threshold B, we expect to find
cases where all inhibitors in the endosperm are off and
all promoters are on, in which case PSGE no longer exists.
The problem of control over seed size is an interesting
one that is not yet solved, not even for a species such as
A. thaliana, which has abundant genetic variation for seed
size (seed size varies a factor 3.5 between accessions [27]).
Understanding the evolution of imprinting thus demands
resolution of this problem.
References
1 Smith, C.C. and Fretwell, S.D. (1974) The optimal balance between the
size and number of offspring. Am. Nat. 108, 499506
2 Geritz, S.A.H. et al. (1999) Evolutionary dynamics of seed size and
seedling competitive ability. Theor. Popul. Biol 55, 324343

www.sciencedirect.com

Vol.12 No.10

443

3 Trivers, R.L. (1974) Parent-offspring conflict. Am. Zool. 14, 249264

4 Haig, D. (2000) The kinship theory of genomic imprinting. Annu. Rev.
Ecol. Syst. 31, 932
5 Burt, A. and Trivers, R. (2006) Genes in Conflict, Harvard University
Press
6 Byers, D.L. et al. (1997) Variation in seed characters in Nemophila
mensziesii: evidence of a genetic basis for maternal effect. Evolution 51,
14451456
7 Fenster, C.B. (1991) Effect of male pollen donor and female seed parent
on allocation of resources to developing seeds and fruit in Chamaecrista
fasciculata (Leguminosae). Am. J. Bot. 78, 1323
8 Alonso-Blanco, C. et al. (1999) Natural allelic variation at seed size loci
in relation to other life history traits of Arabidopsis thaliana. Proc.
Natl. Acad. Sci. U. S. A. 96, 47104717
9 Schruff, M.C. et al. (2005) The AUXIN RESPONSE FACTOR 2 gene of
Arabidopsis kins auxin signalling, cell division, and the size of seeds
and other organs. Development 133, 251261
10 Stokes, D. et al. (2007) Evaluating the utility of Arabidopsis thaliana as
a model for understanding heterosis in hybrid crops. Euphytica 156,
157171
11 Denney, J.O. (1992) Xenia includes metaxenia. Hort. Sci. 27, 722728
12 Wallace, H.M. and Lee, L.S. (1999) Pollen source, fruit set and xenia in
mandarins. J. Hort. Sci. Biotechnol. 74, 8286
13 Seka, D. and Cross, H.Z. (1995) Xenia and maternal effects on maize
kernel development. Crop Sci. 35, 8085
14 Bulant, C. (2000) Xenia effects in maize with normal endosperm: II.
Kernel growth and enzyme activities during grain filling. Crop Sci. 40,
182189
15 Wingard, S.A. (1927) The immediate effect of cross-pollination on the
size of the bean seed. Genetics 12, 115124
16 Lin, B-Y. (1982) Association between endosperm reduction with
paternal imprinting in maize. Genetics 107, 103105
17 Scott, R.J. et al. (1998) Parent-of-origin effects on seed development in
Arabidopsis thaliana. Development 125, 33293341
18 Grossniklaus, U. et al. (1998) Maternal control of embryogenesis
by MEDEA, a Polycomb group gene in Arabidopsis. Science 280,
446450
19 Kinoshita, T. et al. (1999) Imprinting of the MEDEA polycomb gene in
the Arabidopsis endosperm. Plant Cell 11, 19451952
20 Adams, S. et al. (2000) Parent-of-origin effects on seed development in
Arabidopsis thaliana require DNA methylation. Development 127,
24932502
21 Xiao, W. et al. (2006) Regulation of seed size by hypomethylation of
maternal and paternal genomes. Plant Physiol. 142, 11601168
22 Kohler, C. and Makarevich, G. (2006) Epigenetic mechanisms
governing seed development in plants. EMBO Rep. 7, 12231227
23 Feil, R. and Berger, F. (2007) Convergent evolution of genomic
imprinting in plants and mammals. Trends Gen. 23, 192199
24 Kohler, C. et al. (2005) The Arabidopsis thaliana MEDEA Polycomb
group protein controls expression of PHERES1 by parental imprinting.
Nat. Gen. 37, 2830
25 Delph, L.F. et al. (1998) Why fast-growing pollen tubes give rise to
vigorous progeny: the test of a new mechanism. Proc. R. Soc. Lon.
B 265, 935939
26 Kumar, K. and Das, B. (1996) Studies on xenia in almond (Prunus
dulcis (Miller) D.A. Webb). J. Hort. Sci. 71, 545549
27 Krannitz, P.G. et al. (1991) The effect of genetically based differences in
seed size on seedling survival in Arabidopsis thaliana. Am. J. Bot. 78,
446450
28 de Jong, T.J. et al. (2005) Hamiltons rule, imprinting and parentoffspring conflict over seed mass in partially selfing plants. J. Evol.
Biol. 18, 676682
29 Haig, D. (1992) Genomic imprinting and the theory of parent-offspring
conflict. Sem. Dev. Biol. 3, 153160
30 Hurst, L.D. and McVean, G.T. (1998) Do we understand the evolution of
genomic imprinting? Curr. Opin. Genet. Dev. 8, 701708
31 Hoffmann, M.H. et al. (2003) Flower visitors in a natural population of
Arabidopsis thaliana. Plant Biol. 5, 491494
32 Paul, W. et al. (1992) The isolation and characterisation of the tapetumspecific Arabidopsis thaliana A9 gene. Plant Mol. Biol. 19, 611622