vol. 161, no. 4
the american naturalist
april 2003
The Evolution of Androgenesis
Mark J. McKone* and Stacey L. Halpern†
Department of Biology, Carleton College, Northfield, Minnesota
55057
Submitted May 23, 2002; Accepted September 12, 2002;
Electronically published March 28, 2003
abstract: It is well known that some species produce offspring
carrying only female chromosomes by processes such as apomixis
and parthenogenesis (generically termed “gynogenesis”). There are
also several cases of natural reproduction by androgenesis in which
diploid offspring carry nuclear chromosomes from only the male
parent. We used population genetics models to investigate the conditions for invasion of rare androgenesis alleles and the consequences
of their spread. Our models predict that androgenesis alleles often
spread to fixation. If fixation causes the loss of females or female
function in the population, population extinction occurs. Therefore,
androgenesis alleles represent a new class of selfish genetic elements.
Extinction is more likely in dioecious species than in hermaphrodites.
Within dioecious species, extinction is more likely when androgenesis
occurs via paternal apomixis (vs. fusion or doubling of haploid nuclei) and when females are the heterogametic sex (vs. male heterogamety). The apparent rarity of androgenesis compared to gynogenesis could be because androgenesis is harder to detect and more often
leads to population extinction. Also, there could be greater evolutionary constraints on the origin of mutations for androgenesis. We
suggest characteristics of groups in which further cases of androgenesis are more likely to be found.
Keywords: androgenesis, breeding system evolution, gynogenesis, extinction, selfish genetic element, uniparental reproduction.
There are a variety of reproductive systems in which diploid offspring are produced that carry chromosomes from
only the female parent (Bell 1982; Suomalainen et al.
1987), herein referred to collectively as “gynogenesis.”
(The term “gynogenesis” is used in different ways by different authors. For example, in the literature on some
unisexual fish, gynogenesis is used to describe a repro* Corresponding author; e-mail: [email protected].
†
Present address: Department of Ecology, Evolution, and Behavior, University
of Minnesota, St. Paul, Minnesota 55108.
Am. Nat. 2003. Vol. 161, pp. 641–656. 䉷 2003 by The University of Chicago.
0003-0147/2003/16104-020203$15.00. All rights reserved.
ductive system in which a diploid egg is produced apomictically but requires sperm fusion (without male genetic
contribution) before it begins development (Vrijenhoek
1994). Following Stenseth et al. (1985), we would classify
such reproduction as “pseudogamous gynogenesis” and
retain “gynogenesis” as a more general term.) Well-known
examples of gynogenesis include parthenogenetic animals
such as rotifers (Mark Welch and Meselson 2000) and
whiptail lizards (Cnemidophorus; Moritz et al. 1992) as well
as apomictic plants such as dandelions (Taraxacum spp.)
and hawthorns (Crataegus; Richards 1986). An active empirical and theoretical literature has explored the conditions for the evolution of gynogenesis, and investigations
of the evolutionary fate of gynogenetic lineages have played
a leading role in recent arguments about the adaptive significance of sexual versus asexual reproduction (Lively
1992; Mogie 1992; Hanley et al. 1995; Vrijenhoek and
Pfeiler 1997).
“Androgenesis” is defined as reproduction in which diploid offspring carry nuclear chromosomes from only the
male parent. In contrast to various forms of gynogenesis,
most biologists apparently have not considered androgenesis as even a hypothetical possibility. This neglect could
be partly because the nature of oogamy makes it hard to
imagine how a male gamete could provision a developing
embryo. However, androgenesis occurs in at least three
kinds of organisms from two kingdoms. These natural
examples of androgenesis motivated us to model the evolutionary dynamics of this reproductive system, to consider reasons that androgenesis is apparently rare relative
to gynogenesis, and to suggest circumstances in which to
seek further examples of androgenesis.
Known Cases of Androgenesis
Androgenesis has been demonstrated in three types of organism (table 1). Several species of freshwater clams in the
genus Corbicula (family Corbiculidae) apparently reproduce solely by androgenesis (Komaru et al. 1998; Byrne
et al. 2000; Qiu et al. 2001). The sperm of these species
have the DNA content of somatic cells (Komaru et al.
1997; Komaru and Konishi 1999) and presumably arise
by an ameiotic process. After sperm enter the egg, the
642
The American Naturalist
Table 1: Characteristics of organisms known to reproduce regularly and spontaneously by androgenesis
Organism
Several Corbicula species
Cupressus dupreziana
Bacillus rossius and Bacillus grandii hybrids
Type
Breeding system
Source of diploidy in
androgenesis
Freshwater clam
Long-lived conifer
Stick insect
Hermaphrodite
Hermaphrodite
Dioecious, with XO males
and XX females
Diploid sperm
Diploid pollen
Fusion of haploid sperm
pronuclei in egg
Note: See text for references. In Bacillus, there is also a second type of androgenesis in which offspring are produced via diploid sperm.
entire female nuclear genome is extruded from the zygote
in polar bodies (Komaru et al. 1998, 2000). Based on
available phylogenetic analysis and the distribution of
character states (Konishi et al. 1998; Siripattrawan et al.
2000), androgenesis within the genus Corbicula occurs in
lineages with the derived traits of biflagellate sperm (vs.
uniflagellate ancestors), hermaphroditism (vs. dioecious
ancestors), and brooding of young (vs. nonbrooding ancestors). Considerable variation in ploidy exists among
species and populations of Corbicula, but androgenetic
reproduction is known in diploid, triploid, and tetraploid
forms (Komaru and Konishi 1999; Qiu et al. 2001). Current evidence suggests that the higher ploidy levels arose
after the origin of androgenetic reproduction (Qiu et al.
2001).
Cupressus dupreziana (Saharan cypress tree; family Cupressaceae) is the only plant species known to reproduce
typically by androgenesis (Pichot et al. 2001). The pollen
of C. dupreziana is diploid (Pichot and El Maâtaoui 2000).
Microspores are produced by irregular meiosis (El Maâtaoui and Pichot 2001), which often produces unreduced
pollen. Seeds of C. dupreziana often lack electrophoretic
alleles of the maternal tree (Pichot et al. 2000), so it is
likely that the diploid pollen is the sole source of nuclear
genes in the embryo. Only a small proportion (∼10%) of
C. dupreziana seeds contain a viable embryo (Pichot et al.
1998). At present, C. dupreziana is very rare and occurs
only in a small area of the Sahara Desert in Algeria (Stewart
1969; Pichot et al. 2001). Remarkably, the diploid pollen
of C. dupreziana also can develop successfully within the
ovule of the closely related species Cupressus sempervirens
(Pichot et al. 2001).
Androgenesis also occurs in stick insects in the genus
Bacillus (order Phasmatodea, family Bacillidae), although
current information suggests that it is not the typical
breeding system for any species. In some locations in Sicily,
there are persistent hybrids between Bacillus rossius and
Bacillus grandii (Mantovani and Scali 1992; Mantovani et
al. 1999). Reproduction in the hybrids occurs by several
means, rarely including androgenesis of two kinds. The
first occurs when hybrid females backcross with males of
one of the ancestral species (Mantovani and Scali 1992;
Tinti and Scali 1995, 1996). During androgenesis, the nu-
clear genome of eggs of hybrid females degenerates and
is lost; the diploid offspring carry only paternal alleles
(Mantovani and Scali 1992; Tinti and Scali 1996). The
most likely mechanism to produce such an embryo’s diploid genome is fusion of haploid sperm (Tinti and Scali
1996). After multiple sperm enter the Bacillus egg, two
sperm nuclei apparently fuse to produce the diploid nucleus of the zygote (Tinti and Scali 1996). The mitochondria of the egg remains in the zygote, so the offspring have
entirely maternal mitochondrial genes and entirely paternal nuclear genes (Mantovani et al. 1999). This form of
androgenesis produces both male and female offspring
(Mantovani and Scali 1992; Tinti and Scali 1996). The
second type of androgenesis has been observed when hybrid Bacillus males produce all-male progeny that are genetically identical to the father (Tinti and Scali 1995). Presumably the sperm are diploid in this case. Androgenesis
of both types occurs at low frequency in controlled crosses,
and offspring produced by androgenesis have low survivorship (Mantovani and Scali 1992). Though the evidence
for androgenesis is strongest in crosses involving Bacillus
hybrids, Tinti et al. (1995) suggest that it could occur rarely
in nonhybrids as well.
Spontaneous androgenesis also has been observed as a
rare event in other well-studied organisms. For example,
androgenesis by apomixis occurs spontaneously at low frequency in some controlled crosses in the plant Brassica
napus (Chen and Heneen 1989). In Drosophila melanogaster, androgenesis sometimes occurs in a line that carries
a mutation affecting chromosome disjunction during cell
division (Komma and Endow 1995). The offspring are
produced by a process that doubles the ploidy of a haploid
cell produced by meiosis (Komma and Endow 1995). Such
haploid doubling results in homozygosity at all loci, and
all offspring are female (because YY embryos are not
viable).
In addition to these spontaneous cases, androgenesis is
used commonly in artificial breeding of a variety of multicellular organisms, especially angiosperms and teleost
fish. In angiosperms, it is relatively simple to induce haploid microspores to develop into embryos by various forms
of stress (Touraev et al. 1997; Chupeau et al. 1998). Such
microspore embyrogenesis can result in a haploid spo-
Evolution of Androgenesis
rophyte plant, or alternatively it can result in a diploid
sporophyte after a doubling of the haploid genome (Touraev et al. 1997; Beckert 1998; Pechan and Smýkal 2001).
Doubling of haploids occurs spontaneously at variable but
low frequencies in most plants, but the rate can be greatly
increased by chemical treatment (Hansen and Andersen
1998). In fish, the most common technique to induce
androgenesis (Ihssen et al. 1990; Corley-Smith et al. 1996;
Bercsényi et al. 1998) is to destroy the maternal genome
in the egg by irradiation, to allow fertilization by haploid
sperm, and to prevent the first mitotic division of the
zygote by a stress such as heat shock or high pressure. The
result is a diploid embryo from a doubled, haploid, paternal genome.
Models
There has been no formal exploration of the evolutionary
dynamics of androgenesis. Here we use population genetics models to investigate the conditions under which
androgenesis could invade a nonandrogenetic population
and the consequences of the spread of androgenesis alleles.
In particular, we consider how the evolution of androgenesis would be affected by whether a species is hermaphroditic or dioecious, whether androgenesis alleles are
dominant or recessive, different means of achieving offspring diploidy through the male line (apomixis, fusion
of haploids, doubling of haploids), and heterogametic male
(XY) versus heterogametic female (ZW) systems of sex
determination in dioecious species.
We had multicellular animals and land plants in mind
as we developed the models. We generically refer to production of diploid androgenetic sperm in both plants and
animals, although it is likely that the other nuclei of the
male gametophyte would be diploid in androgenetic
plants. The nature of inheritance of androgenesis is not
known for the three established cases, but parthenogenesis
can be determined by a single allele or only a few alleles
(Richards 1986; Mogie 1992; Grossniklaus et al. 2001), so
single-gene models are appropriate.
In contrast to sperm-independent gynogenesis, the
known cases suggest that the process of androgenesis requires either eggs or ovules to provide the resources for
developing offspring produced by androgenesis. The diploid pollen of Cupressus dupreziana develops within female
cones, and the offspring produced by androgenesis in both
Corbicula and Bacillus develop within eggs after loss of the
maternal genome. Thus, our models assume that female
resources limit the capacity for androgenetic reproduction
within a population.
To determine the fate of androgenesis alleles, it is necessary to know the impact of these alleles on fitness of the
carriers. For male fitness, we assigned a value of 1.0 to
643
average male fitness of nonandrogenetic phenotypes and
defined wm to be the average male fitness of androgenetic
individuals. (An “androgenetic individual” or “androgen”
herein refers to an individual that reproduces by means
of androgenesis; we refer to the resulting offspring as “produced by androgenesis.”) We assume that the fitness of
offspring produced by androgenesis will be determined by
their genotype, not their means of production. We know
of no measurements of wm in androgenetic species, so we
determined the effect of varying values on success of androgenesis alleles. Some mechanisms of achieving androgenesis could disrupt meiosis or cause inbreeding, which
could lead to wm ! 1. Other circumstances could produce
wm 1 1. For instance, if androgenesis disrupts or prevents
female reproduction in a hermaphrodite, resources might
be reallocated from female to male reproduction and thus
increase male reproductive success in androgenetic individuals. Analogously, reallocation to female function has
been proposed to occur in male-sterile plants compared
to conspecific hermaphrodites (Darwin 1877; Lewis 1941),
a pattern that has been observed in the male-sterile forms
of a number of gynodioecious species (Delph 1990; Ashman 1994; Poot 1997).
In hermaphrodites, it is important also to consider how
female fitness is affected by the androgenetic phenotype
because androgenetic individuals could also pass on genes
for androgenesis via female reproduction. In plants, gynogenesis has a range of effects on pollen production of
hermaphrodites. Because it is common for apomictic gynogenetic plants to undergo typical meiosis in pollen
production (reviewed by Mogie 1992), we assumed that
female reproduction occurs by normal meiosis in
androgenetic individuals.
We assigned a value of 1.0 to average female fitness of
nonandrogenetic individuals and defined wf to be the average female fitness of androgenetic individuals. It is not
known how the ability to produce offspring by androgenesis would affect the female reproductive success of hermaphrodites. Because of possible disruptive effects of androgenesis mutations on female reproduction, we assume
that it would be unlikely for androgenesis to increase wf
above 1, and so we consider only circumstances where
wf ≤ 1.0. Counterintuitively, wf can be a factor in the evolution of androgenesis even in dioecious species. This is
because some mechanisms of androgenesis can produce
female individuals that carry androgenesis alleles (table 2).
In each case below, we determined the values of wm and
wf that would allow rare androgenesis alleles to invade a
population. Once these invasion criteria were known, we
used the rules of inheritance and simple iteration algorithms (fig. 1; table 2) in an Excel spreadsheet to determine
the fate of invading alleles once they begin to increase in
frequency.
644
The American Naturalist
Table 2: Offspring genotypes produced by androgenetic males that are heterozygous for a dominant androgenesis allele
designated A (parental androgen genotype AAXY)
Mechanism of androgenesis
Apomixis
Fusion of haploid nuclei after meiosis
Doubling of haploid genome after meiosis
Offspring genotypes
(% among nonlethal types)
AAXY (100)
AAXX (8.3)
AAXX (16.7)
AA XX (8.3)
AAXY (16.7)
AAXY (33.3)
AA XY (16.7)
AAYY (lethal)
AAYY (lethal)
AA YY (lethal)
AAXX (50)
AA XX (50)
AAYY (lethal)
AA YY (lethal)
Offspring sex ratio
Offspring that carry A
100% male
33.3% female
100%
75%
66.7% male
…
100% female
50%
…
Note: If A were recessive, only AA individuals would reproduce by androgenesis, and these would produce only AA offspring. Sex ratios
produced by androgenetic males homozygous for A would be as shown in the table, regardless of whether A were dominant or recessive.
Hermaphrodites
We initially consider the case of androgenesis by apomixis
because this is the type known in both Corbicula spp. and
C. dupreziana. In hermaphrodites (including all cosexual
plants), androgenesis alleles can spread via either male or
female reproduction. To determine the conditions for successful invasion by androgenesis alleles, we seek the conditions under which the sum of male and female fitness
of an androgenetic individual exceeds the sum of male and
female fitness of a nonandrogenetic individual.
Suppose that A designates an invading androgenesis
allele and A designates an allele for normal nonandrogenetic reproduction. Because male reproduction occurs
without meiosis, each offspring produced by androgenesis
carries twice the number of paternal genes expected during
sexual reproduction. For example, if A is recessive and
rare, each offspring produced via male reproduction by
an AA individual carries two A alleles (fig. 1). This is
twice the number of A alleles expected under Mendelian
inheritance when A is rare. Similarly, if A is dominant,
each offspring produced via male reproduction of an AA
individual carries one A allele. Under Mendelian inheritance, only half the offspring of an AA parent would carry
the A allele when A is rare.
Thus, the male fitness of an androgenetic hermaphrodite
is twice the number of offspring produced by male reproduction (2wm). The female fitness of an androgenetic
individual (via normal meiosis and production of haploid
eggs) is wf . The total fitness of individual androgenetic
hermaphrodites then is 2wm ⫹ wf . Nonandrogenetic individuals have a male fitness of 1 (by definition) and a
female fitness of 1 (by definition), so the fitness of an-
drogenetic individuals must exceed 2 for successful invasion. These conditions can be written as
wm ⫹
()
wf
1 1.
2
(1)
It is possible that androgenesis in hermaphrodites could
occur by means other than apomixis (see “Fusion of Haploids” and “Doubling of Haploids”) Although such mechanisms produce offspring that are not identical to the paternal parent, the invasion criteria (eq. [1]) are the same
as for apomixis. This follows from the essential fact that
all mechanisms of androgenesis result in androgenetic hermaphrodites having a male fitness of 2wm (offspring carry
twice the number of paternal genes expected under Mendelian inheritance) and a female fitness of wf .
Outcome of invasion by androgenesis alleles in hermaphrodites. If the combination of wm and wf allows invasion
by androgenesis alleles (eq. [1]; fig. 2A), simulations show
that they spread to fixation (fig. 2B, 2C). The outcome is
the same whether the androgenesis allele is dominant or
recessive, although dominant alleles increase in frequency
more rapidly. If the androgenesis phenotype has little effect
on female fitness (wf near 1), androgenesis alleles will
spread to fixation even if male reproductive success is
greatly reduced (values of wm approaching 0.50). If androgenesis completely disrupts female reproduction
(wf p 0), androgenesis alleles still spread to fixation if
there is some compensation in sperm production
(wm 1 1).
Under some circumstances, the fixation of androgenesis
alleles is likely to cause population extinction. At point II
Evolution of Androgenesis
1
wm 1 .
2
645
(2)
Note that the fate of androgenesis alleles is not affected
by wf . Simulations show that androgenesis alleles will
spread to fixation as long as wm 1 1/2, with rate of fixation
faster when the invading androgenesis allele is dominant
than when it is recessive (fig. 3). The number of females
in the population will decline to zero as the allele spreads
(fig. 3B, 3C). Because the androgenetic males cannot reproduce without females, extinction of the population will
occur when the androgenesis genotype spreads to fixation.
Figure 1: Patterns of gamete production and inheritance for a recessive
androgenesis allele A in a hermaphrodite. AA and AA individuals produce haploid sperm or pollen by typical meiosis. AA androgenetic individuals produce diploid sperm or pollen without meiosis. Maternal
reproduction is assumed to be meiotic for all genotypes. Androgenetic
individuals have male (wm) and female (wf) fitness assigned relative to
1.0 for nonandrogenetic individuals.
in figure 2A, female fitness is severely reduced in androgenetic individuals, which reduces wf to only 0.05. When
androgenesis spreads to fixation in the population (fig.
2C), all individuals in the population have this reduced
female fitness and average offspring production drops by
95%. Such a severe drop in reproduction would certainly
increase the likelihood of extinction. If wf p 0, extinction
would be inevitable once the androgenesis alleles spread
to fixation.
Dioecy
Apomixis. When sexes are separate (dioecy), the fate of
androgenesis alleles depends on the mechanism of sex determination. We begin with XY sex determination (male
heterogamety) and consider ZW (female heterogamety)
separately.
In the case of apomictic androgenesis, sperm are produced without meiosis and the offspring of androgenetic
males have the genotype of the paternal parent (table 2).
The androgenetic line acts as an asexual clone within the
population.
Androgenesis alleles will be able to invade as long as
the proportion of males that are androgenetic increases
when the alleles are rare. Androgenetic males produce wm
offspring, all of which are male. Nonandrogenetic males
have a fitness of 1, and half of their offspring are male.
The proportion of androgenetic males in the population
will increase when
Fusion of Haploids. Diploid sperm also could be produced
by fusion of male haploid cells after meiosis, which can
dramatically change the outcome of selection on androgenesis alleles. When there is XY sex determination, random fusion of haploid cells from an androgenetic male
produces XX, XY, and YY cells in the ratio 1 : 2 : 1. Assuming inviability of YY cells, one-third of the offspring
of androgenetic males are females (table 2) that can carry
androgenesis alleles. Thus, unlike the case of apomixis (fig.
3), wf is relevant in determining the outcome of selection.
Androgenesis alleles spread if the average fitness of androgenetic individuals exceeds the average fitness of nonandrogenetic individuals (which is 1 for both males and
females, by definition). Each generation a new cohort of
androgenetic individuals is produced from nonandrogenetic parents. If the androgenesis allele A is recessive, a
cohort of AA individuals is produced each generation by
fusion of A sperm and A eggs from heterozygotes. If A
is dominant, a cohort of AA individuals is produced each
generation when A eggs produced by meiosis in heterozygous females (themselves a product of sperm fusion; see
table 2) are fertilized by A sperm. These cohorts of androgenetic individuals are half male and half female, so
the average fitness of individuals in the cohort is the mean
of the fitness of male and female androgenetic individuals.
Each female in the cohort has a fitness of wf via meiotically produced haploid eggs. Each male in the cohort
produces wm offspring by androgenesis, and these offspring
carry twice the number of paternal genes expected under
Mendelian inheritance. However, the ultimate fitness of
each male in the cohort is not simply 2wm, as is the case
for hermaphrodites. An androgenetic male’s offspring are
two-thirds male and one-third female, and the sexes differ
in fitness; their offspring in turn will be affected by wm
and wf as male and female descendents are produced
through succeeding generations. To determine how these
multiple generations affect the frequency of androgenesis
alleles in the descendents of an androgenetic male, we use
the fact that the frequency of A will increase only if there
646
The American Naturalist
Figure 2: Androgenesis genes in a hermaphrodite population. A, Fate of androgenesis alleles as a function of wm (male fitness of androgenetic
relative to nonandrogenetic individuals) and wf (female fitness of androgenetic relative to nonandrogenetic individuals). B, C, Time course for the
spread of androgenesis alleles and the change in relative number of offspring produced in the population for (B) the fitness values at point I in A
and for (C) the fitness values at point II in A. The androgenesis allele has an initial frequency of 0.05 and is recessive. Once androgenesis is fixed,
the number of offspring produced is reduced because wf ! 1 in both cases. As indicated by density of horizontal lines in A, the risk of population
extinction increases as wf becomes smaller.
is a long-term gain in A alleles to balance the loss of alleles
by production of one-third females by each androgenetic
male in the cohort in the first generation. (A alleles are
reduced in frequency when they are expressed in females
because we assume that wf-1.) The number of alleles lost
via female offspring production is the product of the number of alleles in each androgenetic male’s offspring (2wm),
the proportion of females in those offspring (one-third),
and the gene loss per female (1 ⫺ wf). This product represents the A alleles that must be replaced by subsequent
generations of the male’s descendents to allow invasion
and is added to 2wm to give the total contribution of A
alleles by each androgenetic male in the cohort.
For successful invasion, the mean of female fitness, wf,
and male fitness, 2wm ⫹ [2wm(1 ⫺ wf)]/3, in an androgenetic cohort must exceed 1 (the fitness of nonandrogenetic
individuals). After minor rearrangement the equation is
8wm ⫺ 2wmwf ⫹ 3wf 1 6.
(3)
Outcome of invasion by androgenesis caused by fusion of
haploids. Simulations show that once androgenesis alleles
invade, they rapidly spread to fixation (fig. 4B). The alleles
fix more rapidly when A is dominant. The spread of androgenesis does not necessarily lead to extinction because
androgenetic males always produce one-third female offspring (fig. 4C). As in the hermaphrodite case, extinction
could occur if wf is relatively small. For example, at point
II in figure 4A, wf is 0.05 and wm is 0.85. Once the androgenesis allele has become fixed, the number of offspring
per individual in the population is only 5% of the initial
value (fig. 4D) and extinction could occur because of limited recruitment or demographic stochasticity. If wm 1
0.75, androgenesis alleles spread even when wf p 0; extinction would follow because no offspring would be produced after fixation of A.
Doubling of Haploids. A third mechanism that produces
diploid sperm is the doubling of ploidy within a postmeiotic haploid cell. When there is XY sex determination,
haploid doubling produces half XX sperm and half YY
sperm. Because we assume that YY sperm are inviable, all
offspring produced by androgenetic males are females (ta-
Evolution of Androgenesis
647
Figure 3: Androgenesis alleles in a dioecious population with apomictic production of pollen or sperm. A, Fate of androgenesis alleles as a function
of wm (male fitness of androgenetic relative to nonandrogenetic individuals) and wf (female fitness of androgenetic relative to nonandrogenetic
individuals). B, C, The rate of spread of androgenesis alleles and the change in relative number of offspring produced in the population for the
cases of (B) a recessive A and (C) a dominant A. For the simulations, we assumed the conditions at point I in A (wm p 0.6, wf p 0.1) and an
initial frequency of 0.05 for the androgenesis allele. The sex ratio becomes increasingly male biased as the androgenesis allele spreads (B, C). Extinction
occurs when androgenesis is fixed and no females remain in the population.
ble 2). These females carry androgenesis alleles, and so wf
is important in determining the fate of the invading alleles
in the population.
As in the case of fusion of haploids, we again consider
a cohort of androgenetic individuals newly formed from
nonandrogenetic parents. Half of this cohort is female,
and each female has a fitness of wf . Each male in the cohort
produces wm offspring by androgenesis, and each offspring
carries twice the number of paternal alleles expected under
Mendelian inheritance. But the males in the cohort ultimately do not have a fitness of 2wm because all of their
offspring are females (no XY offspring can result from
haploid doubling; table 2), which each have a fitness of
wf . The individual fitness of each male in the cohort is
therefore 2wmwf . For successful invasion, the average of
the fitness of males (2wmwf) and the fitness of females (wf)
in an androgenetic cohort must exceed 1 (the fitness of
nonandrogenetic individuals). Thus, androgenesis alleles
can invade if
wmwf ⫹
()
wf
1 1.
2
(4)
Outcome of invasion by androgenesis caused by doubling
of haploids. The values of wm and wf that allow invasion
by an androgenesis allele are more restrictive than previous
cases (fig. 5A). Androgenesis only succeeds with relatively
high values of wf , which makes population extinction unlikely. Moreover, simulations show that even when conditions allow invasion, the androgenesis alleles do not
spread to fixation. Instead, a stable polymorphism occurs
at intermediate frequencies of A and A (fig. 5B). Because
androgenetic males produce only female offspring, the sex
ratio becomes female biased as the androgenesis allele
spreads (fig. 5C). The increased proportion of females in
turn can increase per capita offspring production even
when wf ! 1.0 (fig. 5C).
ZW Sex Determination. When females are the heterogametic sex (ZW), the effects of androgenesis on the sex
ratio do not depend on the mechanism of production of
diploid sperm (apomixis, fusion of haploids, or doubling
of haploids). This is because diploid sperm are always ZZ,
and so androgenetic males produce only male offspring.
There is not the added complexity of production of females
648
The American Naturalist
Figure 4: Androgenesis alleles in a dioecious population in which diploidy is achieved by fusion of postmeiotic haploid sperm or pollen nuclei. A,
Fate of androgenesis alleles as a function of wm (male fitness of androgenetic relative to nonandrogenetic individuals) and wf (female fitness of
androgenetic relative to nonandrogenetic individuals). B, Time course for the spread of androgenesis alleles for the fitness values at points I
(wm p 0.65, wf p 0.9) and II (wm p 0.85, wf p 0.05) in A. The androgenesis allele has an initial frequency of 0.05 and is recessive. C, Change in
sex ratio (as proportion females in the population) over time for the fitness values at points I and II in A. The sex ratio becomes only slightly male
biased as the allele spreads because androgenesis by fusion produces one-third female offspring (table 2). D, Offspring production by the population
over time for the fitness values at points I and II in A. The reduction in offspring production depends on the value of wf . As indicated by density
of horizontal lines in A, the risk of population extinction increases as wf becomes smaller.
by androgenetic males as occurs under male heterogamety
in the cases of fusion of haploids or haploid doubling.
No matter what the mechanism of androgenesis, homogametic androgenetic males produce only male offspring that each have a fitness of wm. Nonandrogenetic
individuals produce only half male offspring that each have
a fitness of 1. So the proportion of males that are androgenetic will increase as long as wm 1 1/2 (fig. 3).
Whichever mechanism produces diploid sperm in dioecious species with female heterogamety, the fixation of
androgenesis alleles causes extinction because no females
then remain in the population.
Discussion
If androgenesis mutations arise in a population, our models predict that they will often spread rapidly to fixation.
Fixation of androgenesis alleles would cause immediate
extinction of a population under some circumstances, and
therefore, such alleles represent a new class of selfish genetic elements (Werren et al. 1988). Other selfish genetic
elements that can spread within populations but then
cause extinction include sex chromosomes with meiotic
drive (Hamilton 1967; Lyttle 1977; Jaenike 2001) and cytoplasmic male sterility (Lewis 1941; Frank 1989; Couvet
et al. 1998).
In all cases we analyzed, the fitness values that allow
successful invasion by an androgenesis allele are the same
whether the allele is dominant or recessive. Dominant alleles always spread more quickly (cf. fig. 3A with 3B), as
expected.
The fate of androgenesis alleles is always affected by
relative male fitness (wm) of androgenetic individuals and
Evolution of Androgenesis
649
Figure 5: Androgenesis alleles in a dioecious population in which diploidy is achieved by postmeiotic doubling of haploid sperm or pollen nuclei.
A, Fate of androgenesis alleles as a function of wm (male fitness of androgenetic relative to nonandrogenetic individuals) and wf (female fitness of
androgenetic relative to nonandrogenetic individuals). B, Time course for the spread of an androgenesis allele when at wm p wf p 0.9 (point I in
A). The androgenesis allele has an initial frequency of 0.05 and is recessive. A is able to invade but does not spread to fixation. C, Change in sex
ratio (as proportion females in the population) and in offspring production by the population as A increases in frequency. In this case the sex ratio
becomes more female biased as the androgenesis allele spreads because androgenesis by doubling produces only female offspring (table 2). As a
result of the greater overall proportion of females, offspring production in the population increases as androgenesis spreads even though wf ! 1.
is sometimes affected by relative female fitness of androgenetic individuals (wf). Together, these two parameters
predict the conditions for successful invasion of androgenesis alleles into a population
The process of androgenesis might reduce wm because
of reduced sperm production. In apomictic androgenesis,
a cell that would produce four sperm by complete meiosis
might produce only two if a meiotic division is suppressed.
This is analogous to the reduction in spore production in
some asexual ferns relative to sexual forms (Mogie 1992).
If androgenesis produces offspring by either fusion or doubling of haploid cells in dioecious species, a proportion
of sperm (25% for fusion, 50% for doubling) will be YY
(table 2) and therefore inviable. The effect of these losses
during sperm production on wm will depend on the reproductive biology of the organism. For example, if females typically receive sperm from only one male, a reduction in sperm number may have little impact on
realized male fitness.
Androgenesis could also change wm if dispersal of male
gametes or gametophytes were affected. The diploid pollen
of Cupressus dupreziana is larger than the haploid pollen
of closely related nonandrogenetic Cupressus species (Pichot and El Maâtaoui 2000); a positive correlation between
pollen size and ploidy level is common in angiosperms
(Muller 1979; Bretagnolle and Thompson 1995). Because
C. dupreziana is wind pollinated, the larger size of diploid
pollen could affect its range of dispersal and potentially
reduce wm. Such potential reductions in wm may not necessarily prevent the spread of androgenesis alleles, however,
because androgenesis can be successful even for relatively
small values of both wm and wf .
In addition to its importance in predicting the success
of androgenesis alleles, wf strongly affects the risk of extinction once androgenesis spreads to fixation. In hermaphrodites, extinction will not necessarily occur in androgenetic populations because all individuals retain
female reproductive capacity. However, a low value of wf
could make an androgenetic population vulnerable to extinction once androgenesis was fixed. Two of the three
650
The American Naturalist
known androgenesis cases occur in hermaphrodites, but
apparently with different effects on female reproduction.
In C. dupreziana, seed production is very low on the androgenetic trees (Pichot et al. 1998), and the population
seems to be on the brink of extinction (Stewart 1969;
Pichot et al. 2001). It is possible that C. dupreziana falls
near point II of figure 2A so that androgenesis alleles could
spread despite a large reduction in female fitness. Disruption of female meiosis in C. dupreziana is suggested by
the diploid (and higher) DNA content found in its endosperm (Pichot et al. 1998), which is normally haploid
in conifers because it is derived from the female gametophyte. It would be instructive to have quantitative
estimates of wm and wf in C. dupreziana, perhaps by
estimating components of male fitness (e.g., pollen production per tree, pollen viability, pollen dispersal) and
female fitness (e.g., seed production per tree, seed size,
seedling growth rate) in comparison to closely related nonandrogenetic Cupressus species. Because wf is apparently
so low, we predict that wm would be relatively high.
The situation appears quite different in Corbicula. Several of the androgenetic species have become invasive
worldwide (McMahon 1983), which suggests that there is
relatively little reduction in female reproductive success in
these androgenetic hermaphrodites. So Corbicula could fall
nearer to point I of figure 2A, where androgenesis successfully invades with only a slight reduction in female
reproduction. As in Cupressus, it would be useful to estimate wm and wf by comparing components of male and
female reproductive success between closely related androgenetic and nonandrogenetic forms.
In dioecious species, the spread of androgenesis alleles
is more likely to cause extinction. When androgenesis occurs by apomixis, the increase in frequency of androgenesis
alleles inevitably is accompanied by a decrease in the number of females in the population (fig. 3). The androgenesis
alleles act as sex-ratio distorters because apomictic androgenetic reproduction produces exclusively male offspring.
Once androgenesis is fixed in the population, there are no
females, and reproduction by androgenesis is no longer
possible.
Extinction is less likely when androgenesis is not apomictic. Androgenesis via fusion of haploid sperm only
leads to extinction when wf is severely reduced (fig. 4A).
Androgenesis apparently occurs by fusion of haploid
sperm in the stick insect Bacillus, the only case of androgenesis currently known in a dioecious species. Bacillus
has heterogametic males (though XO rather than XY), and
offspring of both sexes are produced after androgenesis
(Tinti and Scali 1995). So far androgenesis does not seem
to have spread to fixation in a Bacillus species but occurs
only rarely when hybrid females are crossed to a male from
one of the ancestral species. It may be that wf and wm are
too low to allow androgenesis to spread more widely in
Bacillus.
Haploid doubling seems to be the type of androgenesis
least likely to invade a population successfully. First, the
values of wm and wf necessary for successful invasion are
higher than for the other types of androgenesis (fig. 5A).
In addition, the process of haploid doubling produces a
diploid that is homozygous at all loci, so severe inbreeding
depression would result from the expression of any deleterious recessive alleles present in the genome (Suomalainen et al. 1987; Komma and Endow 1995; Corley-Smith
and Brandhorst 1999). With haploid doubling, androgenesis alleles do not spread to fixation even when they are
able to invade (fig. 5B).
Unlike the situation in species with XY sex determination, the source of diploidy (apomixis, fusion of haploids, haploid doubling) has no effect on the evolution of
androgenesis in species with ZW sex determination. Extinction is inevitable if fixation occurs, which suggests that
androgenesis poses a greater risk of extinction should it
arise in species with ZW sex determination, such as birds,
insects in the order Lepidoptera, and a variety of other
animals and plants (Bull 1983).
Comparison with Gynogenesis
Our conclusions about the conditions for spread of androgenesis alleles have some similarities to the predictions
for gynogenetic apomixis. For example, androgenetic
apomixis can invade a dioecious species whenever
wm 1 1/2 (fig. 3), which is analogous to the classical conclusion that all-female asexual genotypes will have twice
the reproductive capacity of sexual forms (Williams 1975;
Maynard Smith 1978). Gynogenetic apomixis alleles will
spread rapidly in hermaphroditic populations as well as
long as the fitness costs are not severe (Lloyd 1977;
Charlesworth 1980; Marshall and Brown 1981). Mogie
(1992) modeled the spread of gynogenetic apomixis alleles
in hermaphrodite populations with consideration of how
such alleles would affect both female ( f ) and male (m)
fitness relative to sexual forms. His conclusion that asexual
alleles will spread if f ⫹ (m/2) 1 1 (Mogie 1992, p. 76) is
directly comparable to our conclusion that androgenesis
alleles can invade if wm ⫹ (wf /2) 1 1.
A critical difference between the evolution of androgenesis and gynogenesis is the risk of immediate extinction. Gynogenesis generally does not require the presence
of males, so reproduction can still occur in populations
fixed for gynogenesis alleles. In contrast, lack of females
will cause populations fixed for androgenesis to become
extinct in several of the cases we modeled. Pseudogamy
(when contact with sperm or pollen is necessary to trigger
gynogenesis but no male genes appear in offspring) is the
Evolution of Androgenesis
type of gynogenesis with evolutionary consequences most
similar to androgenesis. Pseudogamous females cannot reproduce without the presence of males in the population
just as androgenetic males cannot reproduce without females in the population. If pseudogamy became fixed in
a population of a dioecious species, extinction would follow (Stenseth et al. 1985). Pseudogamous populations can
survive in hermaphroditic species because male reproductive capacity is retained in all individuals (Noirot et
al. 1997).
If offspring are produced apomictically, the evolution
of either gynogenesis or androgenesis within a population
changes the type of reproduction from sexual to asexual.
The evolutionary consequences of asexual versus sexual
reproduction are much debated, but a growing body of
theory and evidence suggests that there are circumstances
in which asexual forms would have lower long-term survival (e.g., Gabriel and Bürger 2000; Lively and Dybdahl
2000). The persistence of apomictic forms, whether gynogenetic or androgenetic, would depend on the importance
of these negative effects of asexual reproduction. Such consequences are not explicitly included in our models but
ultimately could cause the loss of apomictic androgenesis.
Even nonapomictic forms of gynogenesis and androgenesis
(such as fusion of same-sex gametes or haploid doubling)
would greatly limit genetic variation of offspring relative
to sexual reproduction, which would have consequences
comparable to obligate selfing.
Why Is Androgenesis Apparently Rarer
than Gynogenesis?
Our models suggest that androgenesis alleles would be
successful under a variety of conditions that seem realistic
for at least some organisms and not very different from
the conditions for success of gynogenesis. Yet gynogenesis
occurs in many different clades of plants and animals (Bell
1982; Richards 1986; Suomalainen et al. 1987), and androgenesis is known from only three. There are a number
of potential reasons for this disparity.
Androgenesis could be harder to detect than gynogenesis. Unless the species is pseudogamous, potential cases
of gynogenesis can be inferred from the ability of a female
to reproduce in the absence of male gametes. The analogous inference does not work for androgenesis because
androgenetic reproduction is not possible without eggs or
ovules. Genetic or cytological analysis is necessary to detect
androgenesis, as is the case for pseudogamous gynogenesis.
Yet androgenesis has been found much less often than
pseudogamous gynogenesis (Stenseth et al. 1985), so it
would seem that difficulty of detection is not the only
cause for the low number of documented androgenetic
lineages.
651
Another possibility is that androgenesis and gynogenesis
arise at similar rates, but androgenetic lines are lost more
often to extinction. Immediate extinction follows the fixation of androgenesis alleles in several of our model scenarios, but fixation of gynogenesis does not directly cause
extinction. This is a fundamental difference between androgenesis and gynogenesis and could be an important
reason for the difference in their observed frequencies.
Finally, it could be that the mutations necessary to produce an androgenetic genotype are less frequent than mutations for gynogenesis. If mutations for androgenesis are
complex or would only occur in unusual circumstances,
they simply may not have arisen in most organisms. At
least four conditions must arise to produce a functional
androgenetic mutation, and lack of any of these could
prevent the origin of androgenesis.
Diploidy from Paternal Chromosomes. There are two means
to produce diploidy in known cases of androgenesis: apomixis and fusion of sperm. The apomictic production of
diploid sperm or pollen could result from relatively small
or few mutations in meiotic machinery (Richards 1986;
Mogie 1992). The occasional production of diploid pollen
is widespread in angiosperms and is often the result of
heritable mutations in meiotic genes (Bretagnolle and
Thompson 1995; Ramsey and Schemske 1998). Diploid
sperm also arise spontaneously in animals. In humans,
for example, diploid sperm occur at a frequency of
0.12%–0.77% (Shi and Martin 2000). Hybrid minnows in
the genus Rutilus regularly produce viable diploid sperm
(Alves et al. 1999). Such results suggest that mutations to
produce diploid sperm or pollen without meiosis would
arise relatively frequently.
Diploidy could also result from fusion of sperm within
an egg cell, as occurs in Bacillus. Polyspermy (entrance of
multiple sperm into the egg) would be a precondition for
such fusion. In taxa such as mammals, there is a block to
polyspermy that acts at the level of the plasma membrane
or cell surface coat (Jaffe and Gould 1985). Other organisms have “physiological polyspermy,” where multiple
sperm typically enter the egg but there is an intracellular
mechanism to prevent more than one sperm nucleus from
fusing with the egg nucleus (Jaffe and Gould 1985). Physiological polyspermy occurs in some insects, molluscs,
birds, salamanders, reptiles, and sharks (as reviewed by
Rothschild 1954; Jaffe and Gould 1985). Polyspermy has
also been observed in some plants (Vigfússon 1970), although apparently it is not the usual process of fertilization. In species in which multiple sperm enter the egg,
there is at least the possibility of fusion of sperm within
the egg to produce diploid offspring by androgenesis.
Androgenesis by haploid doubling is so far not known
in any wild species but has been found to occur sponta-
652
The American Naturalist
neously in rare circumstances (Komma and Endow 1995).
Because the conditions for spread of androgenesis from
haploid doubling are quite restrictive (as discussed above),
it seems unlikely that natural androgenesis of this type will
be found in wild species.
Embryogenesis without Normal Fertilization. In general, experience with artificial androgenesis suggests that there are
few intrinsic barriers to cells becoming embryonic. In
plants, only mild stress can shift the development of microspores from gametophytic to sporophytic to produce
haploid sporophytes (Chupeau et al. 1998; Smýkal 2000).
There is heritable variation in the tendency to become
androgenetic in maize (Zea mays) and other crop plants
(Beckert 1998). Thus, selection could act to increase the
efficiency of androgenesis once it arises.
Removal of Maternal Chromosomes. If the process of androgenesis occurs by means of fusion with an egg cell, as
it does in both Corbicula and Bacillus, there will be a
maternal set of nuclear chromosomes in the cell with the
two paternal sets. The maternal genome must somehow
be eliminated from the cell. This occurs by extrusion in
the polar bodies in Corbicula (Komaru et al. 2000) and
by degeneration of the egg nucleus in Bacillus (Tinti and
Scali 1996). Mechanisms of elimination of entire parental
genomes have evolved in other circumstances, such as in
process of hybridogenesis in some vertebrates (Dawley
1989) and the psr selfish genetic element in the wasp Nasonia vitripennis (Nur et al. 1988). The molecular control
of these events is not yet understood, so it is difficult to
predict how complex their origin might be.
Elimination of maternal genes is not necessary if androgenesis occurs without fusion with an egg cell. This
seems to be the case in C. dupreziana, in which the diploid
pollen travels to a female cone and presumably can develop
into an embryo without fusion (Pichot et al. 2000).
Retention or Acquisition of Mitochondria and Chloroplasts.
In plants, androgenesis can succeed only if the new cell
line has both mitochondria and chloroplasts. If plant androgenesis occurs via diploid pollen without fusion with
a maternal cell, it seems that the mitochondria and chloroplasts in the embryo would have to be paternally inherited. Maternal inheritance of both organelles is widespread in angiosperms (Mogensen 1996), and it could be
difficult to produce an androgenetic embryo from diploid
pollen if the organelles cannot be paternally transmitted.
Interestingly, the family Cupressaceae is characterized
by paternal inheritance of both mitochondria and chloroplasts (Mogensen 1996), and it may not be a coincidence
that the only known case of androgenesis in plants is a
member of this family. Even in plants with strictly maternal
inheritance of organelles under normal circumstances, the
mechanisms for uniparental inheritance are quite variable
and may affect the likelihood that organelles might be
paternally inherited in an androgenetic genotype. For instance, if the organelles are excluded from the paternal
line during early steps in pollen formation (Mogensen
1996), then they may not be present in an offspring produced by androgenesis. Alternatively, if paternal organelles
are typically excluded from embryos at the time of gamete
fusion (Mogensen 1996), then they may be present in
embryos from pollen produced by androgenesis.
When androgenesis occurs by fusion with an egg cell
(as it does in Corbicula and Bacillus), then the organelles
from the maternal line could be present in the zygote
whether or not the paternal organelles persist. For example, mitochondria are maternally inherited in androgenetic Bacillus (Mantovani et al. 1999). As a result, the
offspring have strictly paternal inheritance of nuclear genes
and strictly maternal inheritance of mitochondrial genes.
Evolutionary Responses to Androgenesis
Responder Alleles. When selfish genetic elements invade
populations, there is strong selection for responder genes
to counter their effects (Hatcher 2000). For instance, there
are responder alleles that counter both meiotic drive (Jaenike 2001; Montchamp-Moreau et al. 2001) and cytoplasmic male sterility (Saumitou-Laprade et al. 1994;
Charlesworth and Laporte 1998; Schnabel and Wise 1998).
If populations with androgenesis also typically have responder alleles that prevent the expression of androgenesis,
detection of androgenesis could be difficult. Hybrids between species and populations could be more likely to have
a selfish allele without a responder; many cases of cytoplasmic male sterility were first found in hybrids (Frank
1989). It is noteworthy that androgenesis in Bacillus was
discovered in a hybrid form, and further examples of androgenesis might be found more easily in other hybrids.
So far no alleles are known that specifically counter
androgenesis. In species with internal fertilization, a potential type of responder allele would be one that allowed
females to detect zygotes or embryos produced by androgenesis and block their successful development. The process of genomic imprinting causes gene expression in the
offspring to be dependent on the parental source of a gene
(Solter 1988). This process occurs in diverse organisms,
including mammals (Bartolomei and Tilghman 1997), angiosperms (Haig and Westoby 1991), and insects (Lloyd
2000). Genomic imprinting could make it possible for a
female to evolve the ability to reject any developing embryos that did not express maternally imprinted genes in
order to prevent development of androgenetically produced offspring.
Evolution of Androgenesis
Reduction of Female Allocation in Androgenetic Hermaphrodites. Once androgenesis alleles spread to fixation in a
hermaphroditic species, female reproductive effort is used
to produce offspring that are genetically unrelated to the
maternal “parent.” A similar situation occurs in gynogenetic hermaphrodites that reproduce by pseudogamous
apomixis; because the only function of sperm then is to
trigger development of the eggs, selection is predicted to
reduce male allocation in pseudogamous species (Kirkendall and Stenseth 1990; Noirot et al. 1997; Weinzierl et al.
1998). Selection for progressively reduced sperm production could eventually cause extinction of a pseudogamous
species once there is insufficient pollen or sperm to trigger
apomictic reproduction (Kirkendall and Stenseth 1990;
Noirot et al. 1997). However, loss of male function might
not occur if there is “selfing” so that pollen or sperm from
the same individual triggers its own female reproductive
effort (Mogie 1992; Noirot et al. 1997).
Similarly, there would be a fitness benefit to reduced
female allocation in androgenetic hermaphrodites if such
a reduction allowed resources to be used for other purposes (including male reproductive effort). Loss of female
function in a population of androgenetic hermaphrodites
could ultimately cause extinction of the population. However, by analogy with pseudogamous gynogenetic species,
selection for reduced female function in androgenetic hermaphrodites could be prevented if androgenetic sperm or
pollen develop within the eggs or ovules of the same individual. Such female investment in an individual’s own
androgenetically produced offspring might occur in either
Corbicula (if sperm usually fertilize eggs of the same clam)
or C. dupreziana (if pollen usually develops within female
cones of the same tree). In general, frequent selfing may
prevent population extinction and stabilize androgenesis
in hermaphrodites.
Undiscovered Androgenesis?
In conclusion, we predict that further cases of androgenesis
will be found once the necessary genetic data on inheritance patterns are collected. Based on the constraints we
have identified and the evolutionary dynamics of androgenesis alleles, new cases of androgenesis are more likely
to be discovered in certain types of organisms. For example, two of the three currently known cases of androgenesis occur in hermaphrodites. Further examples in hermaphrodites are likely because the retention of female
function reduces the probability of immediate extinction
after fixation of androgenesis.
In dioecious species, extinction is predicted if androgenesis invades a species with ZW sex determination (such
as birds and butterflies), so androgenesis is more likely to
be found in species with XY sex determination. Compared
653
to androgenesis by apomixis or haploid doubling, androgenesis by sperm fusion gives the greatest chance of both
being able to invade successfully and to avoid extinction
once it has spread to fixation (cf. figs. 3–5). Fusion of
sperm cells outside the egg is a theoretical possibility but
seems less likely than fusion within an egg after the sperm
nuclei have been prepared for fusion by conversion to
pronuclei. We therefore predict that stable cases of androgenesis will be more likely to be found in lineages with
polyspermy because this allows two sperm pronuclei to be
present in the same egg. For example, the newt Notophthalmus viridescens is regularly polyspermous, and its
eggs will continue to develop with only sperm nuclei if
the egg nucleus is artificially removed after fertilization
(Kaylor 1937; Fankhauser and Moore 1941). For functional androgenesis to arise in such a system would require
only the fusion of sperm nuclei and a mechanism to remove or destroy the maternal chromosomes in the egg.
More information on the molecular mechanism of maternal chromosome loss in Corbicula and Bacillus might
allow better predictions about how such mechanisms
might arise.
In cases where the androgenesis occurs without fusion
with an egg, lack of mitochondria and chloroplasts in pollen or sperm could make it difficult for a successful androgenesis mutation to arise. Cupressus dupreziana evolved
in a family with paternal inheritance of both mitochondria
and chloroplasts. Such paternal inheritance of both organelles is rare in plants but does occur in other conifer
families such as Araucariaceae, Cephalotaxaceae, and Taxodiaceae (Mogensen 1996). These taxa could therefore be
fruitful places to seek further examples of androgenesis.
Acknowledgments
We appreciate comments on earlier versions of the manuscript from J. Bull, L. Delph, M. Mogie, M. Neiman, C.
Pichot, and V. Scali. M.J.M. was partially funded by National Science Foundation (NSF) Research Opportunity
Award DEB-0217593 (supplement to NSF CAREER award
DEB-9983879 to U. Mueller at the University of Texas at
Austin). Carleton College also provided sabbatical support
to M.J.M. during the development of this article.
Literature Cited
Alves, M. J., M. M. Coelho, M. I. Próspero, and M. J.
Collares-Pereira. 1999. Production of fertile unreduced
sperm by hybrid males of the Rutilus alburnoides complex (Teleostei, Cyprinidae): an alternative route to genome tetraploidization in unisexuals. Genetics 151:
277–283.
Ashman, T.-L. 1994. Reproductive allocation in hermaphrodite and female plants of Sidalcea oregana ssp. spicata
654
The American Naturalist
(Malvaceae) using four currencies. American Journal of
Botany 81:433–438.
Bartolomei, M. S., and S. M. Tilghman. 1997. Genomic
imprinting in mammals. Annual Review of Genetics 31:
493–525.
Beckert, M. 1998. Genetic analysis of in vitro androgenetic
response in maize. Pages 24–36 in Y. Chupeau, M. Caboche, and Y. Henry, eds. Androgenesis and haploid
plants. Springer, Berlin.
Bell, G. 1982. The masterpiece of nature: the evolution
and genetics of sexuality. University of California Press,
Berkeley.
Bercsényi, M., I. Magyary, B. Urbányi, L. Orbán, and L.
Horváth. 1998. Hatching out goldfish from common
carp eggs: interspecific androgenesis between two cyprinid species. Genome 41:573–579.
Bretagnolle, F., and J. D. Thompson. 1995. Gametes with
the somatic chromosome number: mechanisms of their
formation and role in the evolution of autopolyploid
plants. New Phytologist 129:1–22.
Bull, J. J. 1983. Evolution of sex-determining mechanisms.
Benjamin/Cummings, Menlo Park, Calif.
Byrne, M., H. Phelps, T. Church, V. Adair, P. Selvakumaraswamy, and J. Potts. 2000. Reproduction and
development of the freshwater clam Corbicula australis
in southeast Australia. Hydrobiologia 418:185–197.
Charlesworth, B. 1980. The cost of sex in relation to mating
system. Journal of Theoretical Biology 84:655–671.
Charlesworth, D., and V. Laporte. 1998. The male-sterility
polymorphism of Silene vulgaris: analysis of genetic data
from two populations and comparison with Thymus
vulgaris. Genetics 150:1267–1282.
Chen, B. Y., and W. K. Heneen. 1989. Evidence for spontaneous diploid androgenesis in Brassica napus L. Sexual
Plant Reproduction 2:15–17.
Chupeau, Y., M. Caboche, and Y. Henry. 1998. Androgenesis and haploid plants. Springer, Berlin.
Corley-Smith, G. E., and B. P. Brandhorst. 1999. Preservation of endangered species and populations: a role for
genome banking, somatic cell cloning, and androgenesis? Molecular Reproduction and Development 53:
363–367.
Corley-Smith, G. E., C. J. Lim, and B. P. Brandhorst. 1996.
Production of androgenetic zebrafish (Danio rerio). Genetics 142:1265–1276.
Couvet, D., O. Ronce, and C. Gliddon. 1998. The maintenance of nucleocytoplasmic polymorphism in a metapopulation: the case of gynodioecy. American Naturalist
152:59–70.
Darwin, C. 1877. The different forms of flowers on
plants of the same species. University of Chicago Press,
Chicago.
Dawley, R. M. 1989. An introduction to unisexual verte-
brates. Pages 1–18 in R. M. Dawley and J. P. Bogart,
eds. Evolution and ecology of unisexual vertebrates.
New York State Museum Bulletin no. 466. Albany, N.Y.
Delph, L. F. 1990. Sex-differential resource allocation patterns in the subdioecious shrub Hebe subalpina. Ecology
71:1342–1351.
El Maâtaoui, M., and C. Pichot. 2001. Microsporogenesis
in the endangered species Cupressus dupreziana A. Camus: evidence for meiotic defects yielding unreduced
and abortive pollen. Planta 213:543–549.
Fankhauser, G., and C. Moore. 1941. Cytological and experimental studies of polyspermy in the newt, Triturus
viridescens. Journal of Morphology 68:387–423.
Frank, S. A. 1989. The evolutionary dynamics of cytoplasmic male sterility. American Naturalist 133:345–376.
Gabriel, W., and R. Bürger. 2000. Fixation of clonal lineages under Muller’s Ratchet. Evolution 54:1116–1125.
Grossniklaus, U., G. A. Nogler, and P. J. van Dijk. 2001.
How to avoid sex: the genetic control of gametophytic
apomixis. Plant Cell 13:1491–1498.
Haig, D., and M. Westoby. 1991. Genomic imprinting in
endosperm: its effect on seed development in crosses
between species, and between different ploidies of the
same species, and its implications for the evolution of
apomixis. Philosophical Transactions of the Royal Society of London B, Biological Sciences 333:1–33.
Hamilton, W. D. 1967. Extraordinary sex ratios. Science
(Washington, D.C.) 156:477–488.
Hanley, K. A., R. N. Fisher, and T. J. Case. 1995. Lower
mite infestations in an asexual gecko compared with its
sexual ancestors. Evolution 49:418–426.
Hansen, N. J. P., and S. B. Andersen. 1998. Efficient production of doubled haploid wheat plants by in vitro
treatment of microspores with trifluralin or APM. Plant
Breeding 117:401–405.
Hatcher, M. J. 2000. Persistence of selfish genetic elements:
population structure and conflict. Trends in Ecology &
Evolution 15:271–277.
Ihssen, P. E., L. R. McKay, I. McMillan, and R. B. Phillips.
1990. Ploidy manipulation and gynogenesis in fishes:
cytogenetic and fisheries applications. Transactions of
the American Fisheries Society 119:698–717.
Jaenike, J. 2001. Sex chromosome meiotic drive. Annual
Review of Ecology and Systematics 32:25–49.
Jaffe, L. A., and M. Gould. 1985. Polyspermy-preventing
mechanisms. Pages 223–250 in C. B. Metz, and A. Monroy, eds. Biology of fertilization. Vol. 3. The fertilization
response of the egg. Academic Press, Orlando, Fla.
Kaylor, C. T. 1937. Experiments on androgenesis in the
newt, Triturus viridescens. Journal of Experimental Zoology 76:375–394.
Kirkendall, L. R., and N. S. Stenseth. 1990. Ecological and
evolutionary stability of sperm-dependent partheno-
Evolution of Androgenesis
genesis: effects of partial niche overlap between sexual
and asexual females. Evolution 44:698–714.
Komaru, A., and K. Konishi. 1999. Non-reductional spermatozoa in three shell color types of the freshwater clam
Corbicula fluminea in Taiwan. Zoological Science 16:
105–108.
Komaru, A., K. Konishi, I. Nakayama, T. Kobayashi, H.
Sakai, and K. Kawamura. 1997. Hermaphroditic freshwater clams in the genus Corbicula produce nonreductional spermatozoa with somatic DNA content.
Biological Bulletin (Woods Hole) 193:320–323.
Komaru, A., T. Kawagishi, and K. Konishi. 1998. Cytological evidence of spontaneous androgenesis in the
freshwater clam Corbicula leana Prime. Development
Genes and Evolution 208:46–50.
Komaru, A., K. Ookubo, and M. Kiyomoto. 2000. All meiotic chromosomes and both centrosomes at spindle pole
in the zygotes discarded as two polar bodies in clam
Corbicula leana: unusual polar body formation observed
by antitubulin immunoflurorescence. Development
Genes and Evolution 210:263–269.
Komma, D. J., and S. A. Endow. 1995. Haploidy and androgenesis in Drosophila. Proceedings of the National
Academy of Sciences of the USA 92:11884–11888.
Konishi, K., K. Kawamura, H. Furuita, and A. Komaru.
1998. Spermatogenesis of the freshwater clam Corbicula
aff. fluminea Müller (Bivalvia: Corbiculidae). Journal of
Shellfish Research 17:185–189.
Lewis, D. 1941. Male sterility in natural populations of
hermaphrodite plants. New Phytologist 40:56–63.
Lively, C. M. 1992. Parthenogenesis in a freshwater snail:
reproductive assurance versus parasitic release. Evolution 46:907–913.
Lively, C. M., and M. F. Dybdahl. 2000. Parasite adaptation
to locally common host genotypes. Nature 405:679–681.
Lloyd, D. G. 1977. Genetic and phenotypic models of natural selection. Journal of Theoretical Biology 69:
543–560.
Lloyd, V. 2000. Parental imprinting in Drosophila. Genetica
109:35–44.
Lyttle, T. W. 1977. Experimental population genetics of
meiotic drive systems. I. Pseudo-Y chromosomal drive
as a means of eliminating cage populations of Drosophila
melanogaster. Genetics 86:413–445.
Mantovani, B., and V. Scali. 1992. Hybridogenesis and
androgenesis in the stick-insect Bacillus rossius-grandii
benazzii (Insecta, Phasmatodea). Evolution 46:783–796.
Mantovani, B., M. Passamonti, and V. Scali. 1999. Genomic evolution in parental and hybrid taxa of the genus Bacillus (Insecta Phasmatodea). Italian Journal of
Zoology 66:265–272.
Mark Welch, D., and M. Meselson. 2000. Evidence for the
evolution of bdelloid rotifers without sexual reproduc-
655
tion or genetic exchange. Science (Washington, D.C.)
288:1211–1215.
Marshall, D. R., and H. D. Brown. 1981. The evolution
of apomixis. Heredity 47:1–15.
Maynard Smith, J. 1978. The evolution of sex. Cambridge
University Press, Cambridge.
McMahon, R. F. 1983. Ecology of an invasive pest bivalve,
Corbicula. Pages 505–561 in W. D. Russell-Hunter, ed.
The mollusca. Vol. 6. Ecology. Academic Press, Orlando,
Fla.
Mogensen, H. L. 1996. The hows and whys of cytoplasmic
inheritance in seed plants. American Journal of Botany
83:383–404.
Mogie, M. 1992. The evolution of asexual reproduction
in plants. Chapman & Hall, London.
Montchamp-Moreau, C., V. Ginhoux, and A. Atlan. 2001.
The Y chromosomes of Drosophila simulans are highly
polymorphic for their ability to suppress sex-ratio drive.
Evolution 55:728–737.
Moritz, C., J. W. Wright, and W. M. Brown. 1992. Mitochondrial DNA analyses and the origin and relative
age of parthenogenetic Cnemidophorus: phylogenetic
constraints on hybrid origins. Evolution 46:184–192.
Muller, J. 1979. Form and function in angiosperm pollen.
Annals of the Missouri Botanical Garden 66:593–632.
Noirot, M., D. Couvet, and S. Hamon. 1997. Main role
of self-pollination rate on reproductive allocations in
pseudogamous apomicts. Theoretical and Applied Genetics 95:479–483.
Nur, U., J. H. Werren, D. G. Eickbush, W. D. Burke, and
T. H. Eickbush. 1988. A “selfish” B chromosome that
enhances its transmission by eliminating the paternal
genome. Science (Washington, D.C.) 240:512–514.
Pechan, P. M., and P. Smýkal. 2001. Androgenesis: affecting
the fate of the male gametophyte. Physiologia Plantarum
111:1–8.
Pichot, C., and M. El Maâtaoui. 2000. Unreduced diploid
nuclei in Cupressus dupreziana A. Camus pollen. Theoretical and Applied Genetics 101:574–579.
Pichot, C., A. Borrut, and M. El Maâtaoui. 1998. Unexpected DNA content in the endosperm of Cupressus
dupreziana A. Camus seeds and its implications in the
reproductive process. Sexual Plant Reproduction 11:
148–152.
Pichot, C., B. Fady, and I. Hochu. 2000. Lack of mother
tree alleles in zymograms of Cupressus dupreziana A.
Camus embryos. Annals of Forest Science 57:17–22.
Pichot, C., M. El Maâtaoui, S. Raddi, and P. Raddi. 2001.
Surrogate mother for endangered Cupressus. Nature
412:39.
Poot, P. 1997. Reproductive allocation and resource compensation in male-sterile and hermaphroditic plants of
656
The American Naturalist
Plantago lanceolata (Plantaginaceae). American Journal
of Botany 84:1256–1265.
Qiu, A., A. Shi, and A. Komaru. 2001. Yellow and brown
shell color morphs of Corbicula fluminea (Bivalvia: Corbiculidae) from Sichuan Province, China, are triploids
and tetraploids. Journal of Shellfish Research 20:
323–328.
Ramsey, J., and D. W. Schemske. 1998. Pathways, mechanisms, and rates of polyploid formation in flowering
plants. Annual Review of Ecology and Systematics 29:
467–501.
Richards, A. J. 1986. Plant breeding systems. Allen & Unwin, London.
Rothschild, L. 1954. Polyspermy. Quarterly Review of Biology 29:332–342.
Saumitou-Laprade, P., J. Cuguen, and P. Vernet. 1994. Cytoplasmic male sterility in plants: molecular evidence
and the nucleocytoplasmic conflict. Trends in Ecology
& Evolution 9:431–435.
Schnabel, P. S., and R. P. Wise. 1998. The cytoplasmic basis
of cytoplasmic male sterility and fertility restoration.
Trends in Plant Science 3:175–180.
Shi, Q., and R. H. Martin. 2000. Spontaneous frequencies
of aneuploid and diploid sperm in 10 normal Chinese
men: assessed by multicolor fluorescence in situ hybridization. Cytogenetics and Cell Genetics 90:79–83.
Siripattrawan, S., J.-K. Park, and D. Ó. Foighil. 2000. Two
lineages of the introduced Asian freshwater clam Corbicula occur in North America. Journal of Molluscan
Studies 66:423–429.
Smýkal, P. 2000. Pollen embryogenesis: the stress mediated
switch from gametophytic to sporophytic development:
current status and future prospects. Biologia Plantarum
43:481–489.
Solter, D. 1988. Differential imprinting and expression of
maternal and paternal genomes. Annual Review of Genetics 22:127–146.
Stenseth, N. C., L. R. Kirkendall, and N. Moran. 1985. On
the evolution of pseudogamy. Evolution 39:294–307.
Stewart, P. 1969. Cupressus dupreziana, threatened conifer
of the Sahara. Biological Conservation 2:10–12.
Suomalainen, E., A. Saura, and J. Lokki. 1987. Cytology
and evolution in parthenogenesis. CRC, Boca Raton,
Fla.
Tinti, F., and V. Scali. 1995. Allozymic and cytological
evidence for hemiclonal, all-paternal, and mosaic offspring of the hybridogenetic stick insects Bacillus rossiusgrandii grandii. Journal of Experimental Zoology 273:
149–159.
———. 1996. Androgenetics and triploids from an interacting parthenogenetic hybrid and its ancestors in stick
insects. Evolution 50:1251–1258.
Tinti, F., B. Mantovani, and V. Scali. 1995. Reproductive
features of homospecific hybridogenetically-derived
stick insects suggest how unisexuals can evolve. Journal
of Evolutionary Biology 8:81–92.
Touraev, A., O. Vicente, and E. Heberle-Bors. 1997. Initiation of microspore embryogenesis by stress. Trends
in Plant Science 2:297–302.
Vigfússon, E. 1970. On polyspermy in the sunflower. Hereditas (Lund) 64:1–52.
Vrijenhoek, R. C. 1994. Unisexual fish: model systems for
studying ecology and evolution. Annual Review of Ecology and Systematics 25:71–96.
Vrijenhoek, R. C., and E. Pfeiler. 1997. Differential survival
of sexual and asexual Poeciliopsis during environmental
stress. Evolution 51:1593–1600.
Weinzierl, R. P., K. Berthold, L. W. Beukeboom, and N.
K. Michiels. 1998. Reduced male allocation in the parthenogenetic hermaphrodite Dugesia polychroa. Evolution 52:109–115.
Werren, J. H., U. Nur, and C.-I. Wu. 1988. Selfish genetic
elements. Trends in Ecology & Evolution 3:297–302.
Williams, G. C. 1975. Sex and evolution. Princeton University Press, Princeton, N.J.
Associate Editor: Curt Lively