Understanding the geographic distributions of apomictic plants: a case for a pluralistic approach

Abstract

Asexual organisms usually have larger, and in the Northern Hemisphere, more northern distributions than their sexual relatives. This phenomenon, called geographical parthenogenesis, has been attributed to predispositions in certain taxa, advantages of polyploidy and/or hybrid origin, advantages of uniparental reproduction, introgression of apomixis into sexuals, niche differentiation of clones, and biotic interactions. Here we focus on the role of uniparental reproduction in colonisation, and the importance of different developmental pathways, i.e. autonomous apomixis which does not require pollination and fertilisation of endosperm nuclei for successful seed set, and pseudogamous apomixis which does. A literature survey suggests that geographical parthenogenesis occurs frequently in species with autonomous apomixis, while the correlation with pseudogamy is poorly documented. However, taxonomic patterns (e.g. predominance of Asteraceae) and also methodological bias may influence estimates of frequencies of geographical parthenogenesis. We demonstrate that a flow cytometric seed screen (FCSS) is a powerful method for assessing pseudogamous vs. autonomous apomixis. We show that population genetic studies provide insights into the genetic diversity of apomicts, but do not give strong support for uniparental reproduction being the only explanation of geographical parthenogenesis. Molecular studies help elucidate the evolutionary and biogeographical history of apomictic complexes, and we conclude that multidisciplinary studies are needed to understand fully the phenomenon of geographical parthenogenesis.

Introduction

Apomixis (here defined as agamospermy, i.e. asexual reproduction via seeds) results frequently in an increased abundance and distribution of apomictic species relative to sexually reproducing related taxa. Since Vandel (1928) coined the term ‘geographical parthenogenesis’ to describe this phenomenon, it has been shown to occur both in animals and plants (Bell 1982; Bierzychudek 1985; Asker and Jerling 1992; Van Dijk 2003; Haag and Ebert 2004; Kearney 2005; Hörandl 2006). For higher plants, Bierzychudek (1985) provided the most comprehensive evaluation of earlier literature on the topic and concluded that apomictic plant groups have (1) larger distributions, often considerably exceeding those of their sexual relatives, (2) tend to range to higher latitudes and altitudes than their sexual relatives, and (3) tend to colonise previously glaciated areas. It is also true that sexual relatives often have distributions centred within much larger ranges of apomictic complexes (Antennaria, Bayer 1990; Paspalum, Urbani 2002; Taraxacum and Chondrilla, Van Dijk 2003; Ranunculus, Hörandl and Paun 2007; earlier studies reviewed by Asker and Jerling 1992). Richards (1997) pointed out that geographical parthenogenesis applies only to taxa with gametophytic apomixis. Sporophytic apomixis, which is frequently observed in multi-seeded tropical plants, has completely different developmental and control mechanisms, and is connected with other ecological and geographical features. This mode of reproduction, and also vegetative propagation, which we regard rather as a matter of growth than of reproduction (see Mogie 1992), will not be discussed further here.

A broad evaluation of hypotheses to explain possible causes of geographical parthenogenesis led Bierzychudek (1985) to conclude that distributional success might be mainly due to advantages of polyploidy, which is a common feature of apomictic plants. This idea is rejected by reviews evaluating sexual polyploid plants, because sexual polyploidy is not correlated with large distribution areas (Stebbins and Dawe 1987; Hörandl 2006). In the North Atlantic area, diploid sexuals, polyploid sexuals, sexual selfers and apomictic microspecies do not differ significantly from each other in size of distribution areas, but apomictic complexes as a whole have significantly larger distribution areas than all other groups. These data suggest that geographical parthenogenesis is indeed more related to apomixis than to any other biological trait, although it is not an obligatory feature of apomixis (Hörandl 2006). Several other non-exclusive hypotheses have been proposed for causing geographical parthenogenesis, relying either on intrinsic features of asexual reproduction, such as uniparental reproduction (Stebbins 1950; Baker 1967; Mogie et al. 2007), interactions between sexuals and apomicts (Mogie 1992; Carillo et al. 2002; Mogie et al. 2007; Van Dijk 2007), or on external factors, such as the production of generalists (general purpose genotypes; Baker and Stebbins 1965; Lynch 1984) vs. specialised genotypes (Vrijenhoek 1984, 1994), or biotic interactions with other organisms (Maynard Smith 1978; Bell 1982; Glesener and Tilman 1987). In a recent review, Hörandl (2006) discussed how the following factors contribute to the phenomenon: (1) geographical origin of apomixis around iceshields of the Pleistocene, in areas with frequent events of hybridisation and polyploidy triggering shifts to apomixis; (2) potential advantages of uniparental reproduction; (3) the role of unidirectional introgression of apomixis into sexuals; (4) the role of niche differentiation of clones, and (5) the role of reduced biotic interactions in colder climates, and concluded that in flowering plants no single factor provides a full explanation.

In plants, asexual reproduction might be expected to be less advantageous than in animals. This is because most plants are hermaphroditic and therefore do not suffer from the ‘cost of sex’ caused by producing male individuals (Maynard Smith 1978); the costs of male function are only that of producing male organs. Resource allocation from male to female reproduction has been observed in rare cases of male sterility (Meirmans et al. 2006), but the majority of apomicts maintain male function for fertilisation of primary endosperm nuclei, which is required for formation of viable seed (pseudogamy). In addition, self-fertilisation in hermaphrodites can outweigh the expected advantages of uniparental reproduction for apomicts, in that both selfing and apomixis avoid the ‘cost of outcrossing’ because obligatory selfers and apomicts can serve as pollen donors of progeny of out-crossers, but not vice versa (Holsinger 2000). Selfing is a frequent phenomenon in plants, and well known as a pre-dominant mode of reproduction of colonisers, such as annual pioneer plants, invasive plants and island endemics (Hörandl 2006).

This review focuses on the potential advantages of uniparental reproduction for colonisation. We compare features and implications of apomixis with selfing under various scenarios, and examine autonomous vs. pseudogamous apomicts in an attempt to outline the potential influence of pollen-independence vs. self-pollination on geographical patterns of distribution. We discuss the necessity of efficient screening methods for determining modes of reproduction and present novel approaches to this methodological problem. Finally, we discuss the consequences of uniparental population genetic structure in a geographical context, and the importance of understanding biogeographical histories.

Uniparental reproduction – a comparison of apomixis and selfing

Apomicts and selfers have often been thought to be better colonisers than sexual outcrossers because of uniparental reproduction (e.g. Stebbins 1950; Baker 1967: ‘Baker’s Law’). Both have the potential to found populations by single individuals. Beyond the colonisation phase, uniparental reproduction might alleviate reduced fecundity resulting from pollen limitation in small founder populations, and compensate for lack of suitable pollinators in the colonised region (Rambuda and Johnson 2004). Apomixis has been frequently reported in invasive alien plants (e.g. Heteropogon contortus, Carino and Daehler 1999; Ageratina adenophora, Rambuda and Johnson 2004; Taraxacum officinale, Brock 2004; many species of Hieracium subg. Pilosella, Fehrer et al. 2007; Hypericum perforatum, Barcaccia et al. 2007), supporting the idea that apomixis is advantageous for colonisation. Pollinator-independence may contribute to the abundance of apomicts in colder climates and under conditions of a short growth period, e.g. at higher altitudes and latitudes. In long-term studies on mixed populations, pollination-limited seed set and stronger fluctuations of fecundity in sexual dandelions compared to apomicts was found to be the highest cost of sex (Van Dijk 2007).

Contrary to selfing, apomixis functions without meiosis and fertilisation of egg cells, thus resulting in genetically maternal offspring. Consequently, apomixis infers no increase in homozygosity and avoids inbreeding depression in the offspring. Maintenance of heterozygosity, which is usually higher in apomicts than in related sexual species (e.g. Gornall 1999; Hörandl and Paun 2007), is probably an important advantage to apomicts that frequently experience bottleneck situations or founder events.

In the long term, apomicts are expected to need more time to recover genetic variation after a bottleneck situation than selfers because the latter maintain segregation at heterozygous loci during meiosis (Burt 2000). Asexuals may also suffer a fitness disadvantage because of the accumulation of deleterious mutations (Muller’s ratchet, see Birdsell and Wills (2003). In sexual selfers, selection against homozygous deleterious alleles may lead to purging effects and subsequent increases of fitness after inbreeding (Carr and Dudash 2003). Floristic comparisons of selfers and agamospecies in the North Atlantic area reveal no significant differences in the size of their respective distributions (Hörandl 2006). However, selfing is more frequently observed in island colonisers, which may be due to a higher ability to adapt and speciate in sexuals (Hörandl 2006).

Selfing is frequently associated with polyploidy, which can cause a breakdown of self-incompatibility (s-i) especially in gametophytic s-i systems (Mable 2004; Barringer 2007). Theoretically, polyploidy may reduce rates of homozygote formation (Richards 1997) and can buffer inbreeding depression through masking deleterious mutations by multiple copies of the genome (Lande and Schemske 1985). However, the actual relation of polyploidy to inbreeding depression is more complex and still not fully understood (Barringer 2007). Under the theoretical assumption of relaxed inbreeding depression, self-compatible (s-c) sexual and apomictic polyploids would have rather equal conditions for maintenance, whereas apomicts would hold advantage over self-incompatible sexual polyploids. Compared with diploid, self-incompatible sexuals, apomixis infers a two-fold advantage, i.e. both uniparental reproduction and avoidance of inbreeding depression in small founder populations.

Sexual relatives of apomicts are usually self-incompatible under normal conditions (Asker and Jerling 1992). Nevertheless, experimental studies in Hieracium (Krahulcová et al. 1999; Mráz 2003) and Taraxacum (Morita et al. 1990; Tas and Van Dijk 1999; Brock 2004) have shown that self-incompatibility of sexual taxa can break down when accompanied by additional pollen from another sexual or apomictic species (mentor effect; Richards 1997). Thus, sexuals co-existing with apomicts or other closely related species may reproduce via selfing and avoid introgression by apomicts. A study of mixed populations of native sexual Taraxacum ceratophorum and introduced apomictic T. officinale suggested that genetic assimilation of sexuals partly occurs due to asymmetrical hybridisation, but selfing also partly occurs because of a mentor effect (Brock 2004). Similarly, experimental studies on Ranunculus auricomus suggest mentor effects in diploid sexual individuals after pollination with hexaploid apomicts (EM Temsch and E Hörandl, unpublished). The mentor effect may contribute to maintenance of sexual populations even if apomictic individuals have arisen or invaded within a sexual population.

Autonomous vs. pseudogamous apomixis: taxonomic predispositions or methodological bias?

Apomicts reproduce either independently of pollination (via autonomous apomixis) or via pseudogamy. Pseudogamy is where the embryo develops independently, but pollination is necessary for the fertilisation of endosperm nuclei and formation of germinable seeds (Asker and Jerling 1992; Mogie 1992; Richards 1997; Curtis and Grossniklaus 2008). In pseudogamous apomicts, self-pollination may enable the development of functional seed even without external pollen (observed in Ranunculus auricomus: Rutishauser 1954; Hörandl 2008; and in many apomictic Rosaceae such as Amelanchier, Sorbus, Malus, Crataegus, and Rubus; Nybom 1986; Kollmann et al. 2000; Dickinson et al. 2007). For uniparental reproduction, pseudogamy has the disadvantage of requiring a breakdown of self-incompatibility to allow self-pollen tube growth and endosperm fertilisation. Although this shift in breeding system is frequently observed in polyploids, the correlation is not significant (Mable 2004). In Ranunculus auricomus, autopolyploid sexuals are s-i, whereas s-c is only observed in allopolyploid apomicts. Strong selection for rare s-c genotypes may explain the fixation of s-c in polyploid apomicts (Nogler 1984; Hörandl 2008). Another disadvantage of pseudogamy is that bad pollen quality, a common feature of apomicts due to meiotic disturbances, may impair seed set and female fitness (Hörandl 2008). After crossing between plants of different ploidy level, pseudogamous apomicts may also show seed incompatibility due to shifts of maternal to paternal genome contributions in the endosperm (Nogler 1984). Under these considerations, autonomous apomicts are expected to have better colonising abilities, higher reproductive assurance and consequently higher frequencies of geographical parthenogenesis than pseudogamous apomicts.

An updated literature survey of well-documented cases of geographical parthenogenesis of apomicts vs. sexuals in angiosperms may give further insights in relation to different modes of apomixis (Table 1). This compilation includes only direct comparisons of species with sexual and apomictic populations (sometimes classified as sub-species), and in a second category, apomictic complexes with many apomictic microspecies where relationships of sexual species and the apomictic complex have been assessed. Cases of adventitious embryony (e.g. in orchids) reported in Bierzychudek (1985) and elsewhere were not included. Many apomictic taxa reported in Bierzychudek (1985) do have a distribution in northern or previously glaciated areas, but the closest sexual relatives are unknown; such cases are also excluded from Table 1. This implies some uncertainty for generalisations because many large ‘candidate’ taxa, e.g. the genus Alchemilla (Rosaceae) were not included because of lack of information on sexual species (Fröhner 1990). Altogether only five new well-documented cases of sexual/apomictic pairs of species or subspecies, and four cases of sexual/apomictic complexes could be added to the original survey of Bierzychudek (1985).

Table 1.

Records of geographical parthenogenesis (GP) in species or complexes with both sexual and apomictic cytotypes/microspecies. Mode of apomixis: A, autonomous apomixis; P, pseudogamy (after Gustafsson 1946, Nogler 1984 and the newer literature cited here). Type of GP: L, apomictic distribution range is larger than the sexual one; N, apomicts range further north than the sexual taxa; G, apomicts occur in previously glaciated areas, sexuals do not; no, none of the criteria fulfilled.

Species	Family	Mode of apomixis	GP	Reference
Antennaria friesiana	Asteraceae	A*	L,N	Bayer 1991
Antennaria monocephala	Asteraceae	A*	L,N	Bayer 1991
Antennaria parlinii	Asteraceae	A*	no	Bierzychudek 1985
Arnica alpina	Asteraceae	A	L,N	Bierzychudek 1985; Hufft Kao 2007
Arnica amplexicaulis	Asteraceae	A	L,N	Bierzychudek 1985
Arnica angustifolia	Asteraceae	A	G	Bierzychudek 1985
Arnica chamissonis	Asteraceae	A	L, N	Bierzychudek 1985
Arnica lessingii	Asteraceae	A	L	Bierzychudek 1985
Arnica lonchophylla	Asteraceae	A	L, N, G	Bierzychudek 1985
Arnica longifolia	Asteraceae	A	L, N	Bierzychudek 1985
Arnica louiseana	Asteraceae	A	L, N	Bierzychudek 1985
Chondrilla juncea	Asteraceae	A	L	Van Dijk 2003
Crepis acuminata	Asteraceae	A	no	Bierzychudek 1985
Crepis bakeri	Asteraceae	A	L, N	Bierzychudek 1985
Crepis exilis	Asteraceae	A	L, N	Bierzychudek 1985
Crepis modocensis	Asteraceae	A	L, N, G	Bierzychudek 1985
Crepis monticola	Asteraceae	A	L, N	Bierzychudek 1985
Crepis occidentalis	Asteraceae	A	L, N, G	Bierzychudek 1985
Crepis pleurocarpa	Asteraceae	A	L, N	Bierzychudek 1985
Erigeron strigosus	Asteraceae	A	L, N	Noyes 2007
Eupatorium altissimum	Asteraceae	A	L, N, G	Bierzychudek 1985
Eupatorium cuneifolium	Asteraceae	A	L, N,	Bierzychudek 1985
Eupatorium lecheaefolium	Asteraceae	A	L, N	Bierzychudek 1985
Eupatorium leucolepis	Asteraceae	A	L, N	Bierzychudek 1985
Eupatorium pilosum	Asteraceae	A	L, N	Bierzychudek 1985
Eupatorium rotundifolium	Asteraceae	A	L, N	Bierzychudek 1985
Eupatorium sessilifolium	Asteraceae	A	L, N, G	Bierzychudek 1985
Parthenium argentatum	Asteraceae	P	L, N	Bierzychudek 1985
Townsendia condensata	Asteraceae	A	L, N, G	Bierzychudek 1985
Townsendia exscapa	Asteraceae	A	L, N, G	Bierzychudek 1985
Townsendia grandiflora	Asteraceae	A	no	Bierzychudek 1985
Townsendia hookeri	Asteraceae	A	L, G	Thompson and Whitton 2006
Townsendia incana	Asteraceae	A	L, N	Bierzychudek 1985
Townsendia leptotes	Asteraceae	A	L, N, G	Bierzychudek 1985
Townsendia montana	Asteraceae	A	L, N	Bierzychudek 1985
Townsendia parryi	Asteraceae	A	L, N	Bierzychudek 1985
Townsendia rothrockii	Asteraceae	A	G	Bierzychudek 1985
Townsendia scapigera	Asteraceae	A	no	Bierzychudek 1985
Townsendia spathulata	Asteraceae	A	no	Bierzychudek 1985
Townsendia strigosa	Asteraceae	A	N	Bierzychudek 1985
Boechera holboellii	Brassicaceae	P**	no	Sharbel et al. 2005
Bouteloua curtipendula	Poaceae	P	no	Grant 1981; Bierzychudek 1985
Calamagrostis stricta	Poaceae	A	no	Bierzychudek 1985
Calamagrostis purpurascens	Poaceae	A	no	Bierzychudek 1985
Paspalum simplex	Poaceae	P	L, N	Urbani 2002
Poa cusickii ssp. cusickii	Poaceae	P?*	L, N	Bierzychudek 1985
Poa fendleriana	Poaceae	P?*	L, N	Bierzychudek 1985
Ranunculus kuepferi	Ranunculaceae	P	L, N, G	Huber 1988; Cosendai and Hörandl 2007
Ranunculus parnassifolius	Ranunculaceae	P?	L, N, G	Vuille and Küpfer 1985; Jalas and Suominen 1989
Apomictic complexes
Antennaria rosea complex	Asteraceae	A*	L, N	Bayer 1990
Hieracium pilosella s.l.	Asteraceae	A	N	Bierzychudek 1985; Fehrer et al. 2007
Taraxacum officinale agg.	Asteraceae	A	L, N	Van Dijk 2003
Ranunculus auricomus agg.	Ranunculaceae	P	L, N, G	Hörandl and Paun 2007
Rubus sect. Rubus	Rosaceae	P	L, N	Weber 1995

^*dioecious; uniparental reproduction only possible for autonomous apomixis.

^**also autonomous apomixis reported, but with high frequencies of seed abortion; regarded as effectively pseudogamous.

The low number of cases with pseudogamy compared to that of autonomous apomixis is striking because it is by far the more frequent mode of apomixis (Richards 1997). These data suggest that geographical parthenogenesis may in fact be a feature typical for autonomous apomixis, but because of the low number of cases of pseudogamous apomicts, we refrain here from statistical tests.

Autonomous apomixis has so far been reported almost exclusively in Asteraceae, although it occurs sporadically in e.g. Alchemilla (Rosaceae), in Burmannia (Burmanniaceae), and in some grasses (Calamagrostis and Cortaderia); (Nogler 1984). In Hypericum perforatum and in Boechera holboellii, it occurs rarely, additionally to pre-dominant pseudogamy (Barcaccia et al. 2007; Voigt et al. 2007). In Boechera, autonomous development leads to higher frequencies of seed abortion (Voigt et al. 2007). The reasons for the taxonomically infrequent occurrence of autonomous apomixis are still not well understood, but might be explained by the evolutionary history of angiosperms and functional constraints. Double fertilisation (although not necessarily a triploid endosperm) is probably an ancestral character in the phylogeny of angiosperms (Soltis et al. 2005). In sexual plants, it has the advantage that energy used in the formation of nutritious tissues is invested only after successful pollination and fertilisation of the egg cell has occurred. This advantage may have led to fixation of double fertilisation in the regulatory mechanisms of seed development. Autonomous endosperm development requires a relaxed sensitivity against genomic imprinting in the endosperm, because it leads to a loss of the paternal genome contribution for endosperm development (Koltunow and Grossniklaus 2003). In most angiosperms, the ratio of the maternal to the paternal genome in the endosperm has to remain constant in a 2:1 ratio for normal seed development, whereas the embryo can develop from an unreduced egg cell without a paternal contribution; the endosperm has thus similar features to the placenta in mammals (Koltunow and Grossniklaus 2003; Spielmann et al. 2003; Vinkenoog et al. 2003). In general, higher dosages of the maternal genome lead to under-development, whereas higher dosages of the paternal genome lead to over-proliferation of the endosperm. Unbalanced paternal:maternal genomes in the endosperm is also the main reason for developmental problems of seed in sexual interploidal crosses in angiosperms (Levin 2002), whereby small deviations are sometimes tolerated (see detailed review in Vinkenoog et al. 2003). Autonomous endosperm development is possible either when imprinted genes play an insignificant role in endosperm development in a species, or when both sets of endosperm-inhibiting and endosperm-promoting genes are expressed from the maternally inherited genome. Another possibility is that the role of the endosperm as a source of embryo nutrition is taken over by other tissues, thus avoiding the effects of genomic imprinting (Vinkenoog et al. 2003).

In fact, most pseudogamous plants have developed various mechanisms to maintain the optimal 2 maternal:1 paternal genome endosperm balance. In many apomictic grasses, a four-nucleate embryo sac of the Panicum type is formed, with one unreduced egg cell and one unreduced polar nucleus (2n), which is fertilised by one reduced sperm nucleus (1n), preserving the 2 maternal:1 paternal ratio in the endosperm (Savidan 2007). In many Rosaceae and in Ranunculus auricomus, embryo sacs are 8-nucleate (4n), but both sperm nuclei (2n) are used to fuse with the fused endosperm nuclei, resulting in a 4 maternal:2 paternal genomic ratio of the endosperm (Talent and Dickinson 2007a, b). A similar mechanism works in Dichanthium, only that two unreduced polar nuclei fuse individually with two reduced sperm nuclei. In Boechera holboellii, two unreduced polar nuclei (4n) fuse with one unreduced sperm nucleus (2n), again maintaining the 4 maternal:2 paternal ratio via a modification of pollen development (Voigt et al. 2007).

Autonomous apomixis actually infers a significant advantage with respect to development, because pseudogamy requires other modifications in development to overcome imprinting problems. Moreover, bad pollen quality may reduce seed set in pseudogamous apomicts (Nogler 1984; Hörandl 2008); lower female fitness in apomicts has been also observed in Boechera (Sharbel et al. 2005). In contrast, autonomous apomicts have often higher seed set than their sexual relatives (Michaels and Bazzaz 1986; Van Dijk 2007).

Caution is required when interpreting data on the effects of modes of apomixis on geographical patterns. The survey in Table 1 shows a predominance of geographical parthenogenesis in Asteraceae compared to other families, despite the fact that Poaceae and Rosaceae have higher frequencies of apomictic genera: Rosaceae, 12 apomictic genera/85 genera in total = 14.1% (Dickinson et al. 2007); Poaceae, c. 36/675 = 5.3% (Carman 1997); Asteraceae, 22/1600–1700 = only 1.3–1.4% (Funk et al. 2005; Noyes 2007). These three families contain 75% of all apomictic genera (Carman 1997). The over-representation of Asteraceae in our survey may relate to a predisposition of the family for a relaxed sensitivity to genomic imprinting in the endosperm, as discussed above. So far, pseudogamy has been reported for only one genus in the family, Parthenium; all other apomictic genera of Asteraceae studied show autonomous apomixis (Noyes 2007). In Arnica cordifolia, pseudogamy occurs rarely together with the predominant autonomous development (Hufft Kao 2007). In Taraxacum, embryos may mature despite being accompanied by a heavily retarded endosperm, because the nutritive function of the endosperm seems to be taken over by the nucellus (Vinkenoog et al. 2003). In Asteraceae, the endosperm is in general small, and total endosperm loss may occur during ripening of the embryo and seed storage (e.g. Krahulcová and Suda 2006). A relaxed sensitivity against endosperm imbalance might also infer a tolerance of interploidal crosses in the family, and a higher potential for forming polyploid complexes. More detailed studies on other genera of Asteraceae are missing, and in general most work on genomic imprinting has been made just on the model species Arabidopsis thaliana (Brassicaceae). Another aspect is the general distributional success of the family. Asteraceae have a global distribution (except Antarctica), and members of this family are often important components of ‘at risk’ habitats in islands and other pioneer habitats (Funk et al. 2005).

Another possible bias in the current survey of literature may be introduced by methodological problems. Autonomous apomixis can be often rather easily assessed via simple pollen-exclusion tests: emasculation plus bagging of flowers resulting in normal seed set should give evidence for autonomous apomixis (e.g. Richards 1997). Pseudogamy, in contrast, is much more difficult to prove. The above-mentioned pollen-exclusion test will lead to seed abortion as in any sexual plant. Bagging of flowers without emasculation is only a test for self-compatibility, not for apomixis; results of this test, therefore, cannot discriminate a pseudogamous apomict from a sexual selfer. Pseudogamy can be assessed only by microscopic observation of embryo sac development (clearing techniques, e.g. Herr 1971, 1992), requiring documentation of all stages of early development, high methodological experience and good microscopes. The auxin test for callose deposition is a method so far only used for Poaceae. Progeny tests using DNA fingerprinting or chromosome counts are time-consuming, the former also expensive, and may be biased by other factors such as germination rates, and marker sensitivity (see Leblanc and Mazzuccato 2001). Analysis of genetic diversity of natural populations is only an indirect method but can be informative using codominant markers, e.g. isoenzymes and microsatellites. Such markers may discriminate apomicts from outcrossers and selfers by higher frequencies of heterozygous loci (Hörandl and Paun 2007). Nevertheless, heterozygosity might be influenced by other factors, e.g. strong mutational dynamics of microsatellite loci (Paun and Hörandl 2006). Because of these methodological problems, records of autonomous apomixis might be overrepresented in our literature survey. In fact, geographical parthenogenesis might be expected in widespread facultative pseudogamous apomicts such as Poa pratensis and Hypericum perforatum (Barcaccia et al. 2006, 2007), but detailed distribution maps of apomictic vs. sexual accessions have not yet been published for these taxa. Novel approaches for a rapid and reliable assessment of pseudogamous vs. autonomous apomixis are clearly needed.

Flow cytometric seed screen: a methodological advance in the assessment of modes of reproduction

A major problem for gathering sufficient data for geographical parthenogenesis is the difficulty of obtaining evidence for pseudogamy. In pseudogamous complexes, assessment of apomixis is often based on indirect methods, such as reduced pollen quality and determination of ploidy. These methods incur some uncertainty because pollen may also be partly aborted in hybrids of sexual species, and polyploidy is frequent in sexual angiosperms.

Potential help is offered by flow cytometric seed screen (FCSS), a method developed by Matzk et al. (2000) and recently reviewed by Matzk (2007). It is based on DNA content measurement and estimation of ploidy in embryos vs. endosperm in mature seeds. The ratio of embryo to endosperm DNA content is informative of mode of reproduction, and can also give information on whether the endosperm was fertilised or not. It can thus discriminate between pseudogamy, autonomous apomixis and sexual reproduction (Figure 1). Table 2 provides a summary of expected ratios for simple model examples. The method can also be used to assess partial apomixis, e.g. the formation of dihaploids or polyhaploids (i.e. a polyploid apomictic mother plant forms occasionally a meiotically reduced egg cell that develops parthenogenetically, n+0) or B_III hybrids (an unreduced egg cell that has been fertilised by sperm nuclei, 2n+n; see Matzk 2007). In sexual polyploid complexes, ploidy estimation can be used to rapidly test progeny produced from crosses between different ploidy levels; the method has thus a number of applications in evolutionary studies and in plant breeding (Matzk 2007). The advantage of FCSS is that in apomicts not only the mode of embryo sac formation, but also fertilisation events can be reconstructed in one analytical step, and thus the procedure is more informative than most microscopic embryological studies.

Figure 1.

Scheme of the different developments of seeds in sexual and apomictic plants with the expected ratios of Cx-values of embryo versus endosperm. a, diploid sexual; b, tetraploid sexual; c–d, tetraploid apomicts; c, Pseudogamous I, one pollen nucleus fertilises the endosperm nuclei; d, Pseudogamous II, both pollen nuclei fertilise the endosperm nuclei; e, autonomous endosperm development (no pollen nuclei fertilise the endosperm).

Table 2.

Expected DNA content ratios (expressed as monoploid genome sizes, Cx-values; for terminology, see Greilhuber et al. 2005) of the embryo compared to the endosperm for simple model examples (8-nucleate embryo sac; m, maternal copies; p, paternal genome copies). The pollen donor has the same ploidy level as the mother plant; pollen is meiotically reduced. For further examples, see Matzk et al. (2000) and Matzk (2007).

Mother plant	Embryo (m+p)	Endosperm (m+p)
Diploid sexual (2n=2x)	2Cx (1+1)	3Cx (2+1)
Tetraploid sexual (2n=4x)	4Cx (2+2)	6Cx (4+2)
Tetraploid apomict, autonomous endosperm (2n=4x)	4Cx (4+0)	8Cx (8+0)
Tetraploid apomict, pseudogamous (2n=4x);  one sperm nucleus fuses with endosperm nuclei	4Cx (4+0)	10Cx (8+2)
Tetraploid apomict, pseudogamous (2n=4x);  both sperm nuclei fuse with endosperm nuclei*	4Cx (4+0)	12Cx (8+4)

^*predominant in Ranunculus auricomus and also found in some Rosaceae (Talent and Dickinson 2007a).

Flow cytometric seed screen has been used in various genera (Matzk et al. 2000; Cáceres et al. 2001; Voigt et al. 2007; Barcaccia et al. 2006, 2007; Hufft Kao 2007; Talent and Dickinson 2007a, b; see also Matzk 2007) and is recommended for large samples. Single seeds can be analysed if they have a sufficient size (e.g. Poa pratensis, Triticum aestivum, Zea mays, Matzk et al. 2000; R. carpaticola, see below), while bulked samples will increase the amount of tissue for the analysis of small seeds (e.g. Hypericum, Arabidopsis) and provide an overview for large samples. In the case of facultative apomixis, different progeny classes can be quantified in bulked samples of 10 seeds by the discrimination of the respective peaks of histograms (Krahulcová and Suda 2006). In highly facultative apomicts, such as in Hypericum perforatum or in Hieracium subgen. Pilosella, FCSS discriminated between sexually and apomictically derived progeny, because both the peaks resulting from pseudogamy and from fertilisation appeared in flow histograms (Krahulcová and Suda 2006; Barcaccia et al. 2007). The method is useful for testing progenies in experimental crossings (e.g. Krahulec et al. 2006) and assessing mentor effects (e.g. Mráz 2003). It may also be used for taxa in which progeny tests are difficult to perform because of low germination rates or extended seed dormancy, e.g. for high alpine species. It has been shown that the number of nuclei within seed tissues might decrease after storing, but Hieracium seeds could be analysed successfully with FCSS even after being stored for two years (Krahulcová and Suda 2006).

Here we present two examples of the application of FCSS for rapid assessment of reproductive pathways in plants. In our first example, we used achenes of Ranunculus kuepferi to test if the mode of reproduction of the species is apomictic as suggested by Huber (1988). Ranunculus kuepferi is an alpine plant showing geographical parthenogenesis in the Alps (Cosendai and Hörandl 2007). Apomixis is difficult to assess via clearing techniques because the species often does not flower in cultivation, and progeny tests fail because of very low germination rates (Huber 1988; P. Kuepfer, pers. comm.). The results for two samples are presented here: one was a diploid individual sampled from Col de Rousset in Vercors, France, and the other was a tetraploid from Col du Grand Saint Bernard, Switzerland/Italy. Details of methods of analysis conducted on achenes collected from both plants in the wild will be presented elsewhere. Histograms of the two examples are shown in Figure 2. From the ratios of DNA content of embryo to endosperm we can infer that diploid R. kuepferi shows normal sexual development (Figure 2A), while tetraploid Ranunculus kuepferi is a pseudogamous apomict possessing a ratio of 4:10Cx (Figure 2B), with one pollen nucleus having fertilised the endosperm nuclei (see Figure 1c). Thus FCSS is a very efficient method for evaluating mode of reproduction in this difficult species.

Figure 2.

Flow cytometry histograms (fluorescence intensity vs. number of particles) of Ranunculus kuepferi. A, histogram of six achenes of diploid Ranunculus kuepferi (Rk), individual Vercors_5, showing peaks at the 2Cx (embryo) and the 3Cx (endosperm) ploidy level positions. The three peaks of the internal standard Zea mays (Zm) correspond to 2Cx phase G1, 4Cx phase G2 (nuclei in mitotic divisions) and 8Cx endopolyploid nuclei; B, histogram of six seeds of the tetraploid individual from Grand Saint Bernard_21 showing peaks at the 4Cx (embryo and maternal tissue) and the 10Cx (endosperm) ploidy level positions with the same standard as in A; in both A and B the short tables show the analyses of the gated peak with the number of the peaks, the index ratio of the peak, the mean value of the linear position peak and CV% (coefficient of variation of the peak).

Our second example of the use of FCSS compares diploid and hexaploid plants of the Ranunculus auricomus complex and involves an experimental test of the mentor-effect. A diploid sexual plant of otherwise self-incompatible Ranunculus carpaticola (see Hörandl 2008) was hand-pollinated with pollen from a hexaploid apomictic population. Achenes produced were subjected to FCSS. Figure 3 shows a histogram of a PI stained s-i Ranunculus carpaticola sample of a single seed comprising one 2n (2Cx) peak, which includes the embryo (potentially also some cell nuclei of the testa) and one endosperm peak (3n=3Cx). The fluorescence intensity mean value of the peaks expresses the expected ratio of 2:3 due to selfing caused by the mentor-effect, and excludes the alternative possibility of fertilisation via 3x pollen from a hexaploid plant.

Figure 3.

Flow cytometry histogram (fluorescence intensity vs. number of particles) of seeds of diploid Ranunculus carpaticola showing peaks at the 2Cx (embryo and testa) and the 3Cx (endosperm) ploidy level positions, indicative of induced selfing (mentor-effect, see text). Here only an external standard from maternal somatic tissue was used (histogram not shown).

It is worth noting that DNA content measurement of embryo and endosperm via DNA image cytometry (Vilhar et al. 2001) has in principle the same potentials as FCSS and may give more detailed information, because microscopic observation of dividing nuclei may often allow direct chromosome counting (Hörandl and Greilhuber 2002; J Greilhuber, pers. comm.). However, because of higher preparative demands (the embryo and endosperm must be separated before analysis) the method is less suitable for screening large samples.

Geographical patterns of genetic diversity

Uniparental apomictic reproduction is expected to result in the formation of large clones. For both, autonomous and pseudogamous apomicts, numerous molecular studies have confirmed the existence of populations comprising only one lineage, suggesting that single individuals may have founded such populations (Gornall 1999; Hörandl and Paun 2007). This infers an advantage of apomixis in founder events, although frequency-dependent selection can also lead to the predominance of single clones in populations (Van Dijk 2003). If uniparental reproduction infers better colonising abilities, one would expect that single successful clones might become widespread. However, this is not supported by molecular data, which show that the great majority of apomictic genotypes (92–100%), as assessed by high-sensitive markers, are restricted to single sites (Palacios and Gonzalez-Candelas 1997; Carino and Daehler 1999; Palacios et al. 1999; Van Der Hulst et al. 2000; Houliston and Chapman 2004; Paun et al. 2006a; Cosendai and Hörandl 2007). Widespread apomictic genotypes have been recovered only in studies that have employed a low number of loci, which tend to underestimate clonal diversity (e.g. isozymes), and even then at very low frequencies (Gornall 1999). In general, such lineages show continuous distributions without disjunctions in well-sampled areas (Bayer 1990; Battjes et al. 1992; Kraft et al. 1996; Hörandl et al. 2001). In contrast, the employment of large numbers of dominant markers such as RAPDs or AFLPs suggest that genetic diversity in apomicts is mainly distributed among populations (e.g. Palacios et al. 1999; Paun et al. 2006a), and that isolation-by-distance is not commonly found (see Hörandl 2006; Barcaccia et al. 2006). In conclusion, population genetic data alone provide only equivocal support that uniparental reproduction is the main factor for generating a large distribution range.

The specific biogeographical history of an apomictic group might be another factor shaping geographical patterns (Hörandl 2006). Boechera holboellii, a species widespread in North America, comprises triploid apomicts, and diploid cytodemes that can be sexual or apomictic. Phylogeographic studies based on chloroplast haplotyping and ITS sequence polymorphism have revealed complex biogeographical scenarios, including past fragmentation, recolonisation, isolation-by-distance, but also glacial refugia (Dobeš et al. 2004a, b). Polyploid apomicts predominate in the southern United States, suggesting that apomixis was probably advantageous for colonisation of the southern Sierra Nevada, whereas diploids (sexual or apomictic) prevailed rather in the northern United States (Dobeš et al. 2004b). Analysis of B-chromosomes and degree of pollen abortion, features associated with apomixis in Boechera, revealed no clear geographical pattern of sexuals vs. apomicts and thus confirmed that B. holboellii does not show a typical pattern of geographical parthenogenesis (Sharbel et al. 2005). Multiple formation of polyploids, and frequent hybridisation events have shaped the history of the complex (Koch et al. 2003; Sharbel et al. 2005) and may have resulted in a diffuse geographical pattern. In the Ranunculus carpaticola-cassubicifolius complex, AFLP and microsatellite studies have shown that sexual populations have probably undergone extensive range fragmentation during the Pleistocene, and secondary contact hybridisation has probably led to the origin of hexaploid apomicts (Paun et al. 2006b). Successful niche exploration in human-influenced habitats may have promoted the spread of apomictic populations, resulting in large distributions in the forest belt of Slovakia (Paun et al. 2006a). Here range expansion into these areas is the main component of local geographical parthenogenesis. In the sexual North American cytotypes of Townsendia hookeri, an analysis of chloroplast haplotypes has revealed a scenario of range fragmentation, with refugia north and south of areas glaciated during the Wisconsin glaciation. Polyploid apomicts originated multiple times from sexuals and have colonised previously glaciated areas in the Rocky Mts between refugia harbouring sexual lineages (Thompson and Whitton 2006). In this case, better colonising abilities of devastated, glaciated areas are probably of paramount importance. In summary, these examples demonstrate that the evolutionary and biogeographical history of sexual and apomictic taxa have had a major impact in shaping their respective geographical distributions and therefore need to be studied case by case.

Conclusion – outlines for future research

Geographical parthenogenesis is a complex phenomenon, influenced by both intrinsic and extrinsic factors (Hörandl 2006). To understand its cause, it is first necessary to document the geographical pattern of apomixis and sexual reproduction over the whole distribution range of a species or species complex. Uniparental reproduction, enabling colonisation of areas via single individuals and the provision of reproductive assurance without pollinators, has often been regarded as the most important intrinsic factor contributing to geographical parthenogenesis. Here, however, we suggest that mode of apomixis, i.e. autonomous vs. pseudogamous apomixis, might also be an important determinant of geographical range. Clearly, detailed assessment of mode of apomixis and fitness over the geographical ranges of taxa are required to fully understand the observed patterns of geographical parthenogenesis. In sexual relatives, it will be essential to study the amount of selfing and inbreeding depression that occurs. To this end, flow cytometric seed screen (FCSS) will likely be an important tool in assessing mode of reproduction in apomictic complexes in the future. Population genetic and phylogeographical studies are needed to explore the evolutionary and biogeographical history of apomicts. In summary, a multidisciplinary approach will be necessary for a full understanding to be gained of the complexities of geographical parthenogenesis.