The role of plastids in plant speciation


  • Dedicated to Professor Olga Kulaeva on the occasion of her 80th birthday.

Stephan Greiner, Max-Planck-Institut für Molekulare Pflanzenphysiologie, Wissenschaftspark Golm, Am Mühlenberg 1, D-14476 Potsdam, Golm, Germany. Fax: +49 331 567 87 01; E-mail:


Understanding the molecular basis of how new species arise is a central question and prime challenge in evolutionary biology and includes understanding how genomes diversify. Eukaryotic cells possess an integrated compartmentalized genetic system of endosymbiotic ancestry. The cellular subgenomes in nucleus, mitochondria and plastids communicate in a complex way and co-evolve. The application of hybrid and cybrid technologies, most notably those involving interspecific exchanges of plastid and nuclear genomes, has uncovered a multitude of species-specific nucleo-organelle interactions. Such interactions can result in plastome–genome incompatibilities, which can phenotypically often be recognized as hybrid bleaching, hybrid variegation or disturbance of the sexual phase. The plastid genome, because of its relatively low number of genes, can serve as a valuable tool to investigate the origin of these incompatibilities. In this article, we review progress on understanding how plastome–genome co-evolution contributes to speciation. We genetically classify incompatible phenotypes into four categories. We also summarize genetic, physiological and environmental influence and other possible selection forces acting on plastid–nuclear co-evolution and compare taxa providing molecular access to the underlying loci. It appears that plastome–genome incompatibility can establish hybridization barriers, comparable to the Dobzhansky–Muller model of speciation processes. Evidence suggests that the plastid-mediated hybridization barriers associated with hybrid bleaching primarily arise through modification of components in regulatory networks, rather than of complex, multisubunit structures themselves that are frequent targets.


The genomes of plants, as for all eukaryotes, are compartmentalized, a feature that differs fundamentally from prokaryotic genomes. Eukaryotic genomes, composed of the nucleus and organelles, are the result of endosymbioses, followed by complex reorganization of the genomes of the symbiotic partners (Box 1—reviewed in Herrmann 1997; Herrmann & Westhoff 2001). Plastids and mitochondria have retained remnants of their ancestral genomes, as well as the mechanisms to maintain and express this information. The constitution of organelle genomes differs from that of nuclear genomes, which has profound consequences for function and evolution (Herrmann & Possingham 1980; Bock 2007; Barbrook et al. 2010). Notable features of plastid genomes (plastomes—Renner 1934) are a high redundancy and a virtual lack of recombination; plant cells contain many (in higher plants up to several thousands) identical copies of the plastome (Bock 2007; Rauwolf et al. 2010). During cell division, genetically different plastids in hybrid cells therefore segregate randomly, referred to as somatic segregation and sorting-out (Kirk & Tilney-Bassett 1978; Birky 2001), demonstrated, for example, in variegated tissue (Fig. 1). This also causes the non-Mendelian heredity patterns (Kirk & Tilney-Bassett 1978; Hagemann 2004). The degree of plastome redundancy and the ratio of nuclear and plastid genomes in cells vary markedly with development and appear to be stringently regulated (Herrmann & Possingham 1980; Zoschke et al. 2007; Rauwolf et al. 2010). Today, our understanding of the partite eukaryotic genome is that it functions as an integrated genetic system that is regulated spatiotemporally and quantitatively and co-evolves in a species-specific manner (reviewed in Herrmann 1997; Herrmann & Westhoff 2001).

Figure 1.

 Plastome–genome incompatibility and hybrid variegation in Passiflora and Oenothera. (a, c, d) Hybrid variegation generated by two plastomes transmitted from both sexes to the zygote in F1, of which only one of the plastomes is incompatible with the nuclear genome (dominant plastome–genome incompatibility, see text). Separation of the two homoplastomic tissue types occurs during cell divisions and results from the statistical process of sorting-out of the two plastid types (Kirk & Tilney-Bassett 1978; Birky 2001). (b, e–g) Homoplastomic plants containing only one incompatible plastid type (hybrid bleaching). (a) An F1 individual of Passiflora menispermifolia × P. oerstedii. The incompatible plastid type is from P. menispermifolia. (b) Oe. elata ssp. elata with the foreign plastome III of Oe. glazioviana (genotype AA-III). (c) F1 hybrid of Oe. grandiflora × Oe. oakesiana with plastome III of Oe. grandiflora and plastome IV of Oe. oakesiana (genotype BC-III/IV, white tissue contains plastome III, green tissue plastome IV). (d) Oe. grandiflora with native plastome III and foreign plastome I of Oe. elata ssp. elata (genotype BB-III/I, white tissue contains plastome I, green tissue plastome III). (e) Oe. grandiflora with foreign plastome IV of Oe. parviflora (genotype BB-IV). (f) Oenothera biennis‘mut. lutescens’ (genotype BB-II). (g) F1 generation of Oe. elata ssp. elata × Oe. biennis (genotype AB-I). Phenotypes: (a, c, d) albina or xantha; (b, e) virescent; (f, g) lutescent (cf. Fig. 3).

Co-evolution of the cellular subgenomes is evident from the consequences of interspecific exchanges of plastids and nuclei. Even between closely related species, such an exchange can cause severe developmental disturbances of the resulting plastome-genome hybrid and, in extreme cases, even culminate in lethality (e.g. Stubbe 1989; Kushnir et al. 1991; Herrmann et al. 2003; Levin 2003; Schmitz-Linneweber et al. 2005). Interspecific plastome-genome incompatibilities have been widely observed (Grun 1976; Kirk & Tilney-Bassett 1978), and their implications for speciation were recognized already in the early period of formal genetics (Renner 1934; Stebbins 1950; Stubbe 1964). Molecular data indicate an enormous degree of interdependence between the cellular subgenomes, highly complex gene expression in and between organelles, and a much greater influence of plastomes on ontogenesis and speciation of plants than their mere coding potential suggests (usually <1% of the cellular gene pool for organelle genomes in general) (Kirk & Tilney-Bassett 1978; Bock 2007; Barbrook et al. 2010). Compartmental co-evolution, the topic of this article, has emerged as an important element of reproductive isolation that defines species, and represents one of three basic processes that shaped eukaryotic genomes (Herrmann & Westhoff 2001; Herrmann et al. 2003). The significant developmental impact and limited coding potential of plastomes, combined with organelle genetics which is a robust experimental discipline, provide tangible, causal approaches to understanding evolution of diversity. In this review, we summarize the principal aspects of compartmental co-evolution and their impact on speciation. We assess the findings and limitations of current studies and discuss future prospects, including appropriate case-study models for which key cytological, geographical and phylogenetic contexts are well established.

Box 1. General aspects of plant genome evolution

Cellular cohabitation that led to eukaryotic organisms was a major force in the development of global life. It was only the eukaryotic cell type that permitted the development of advanced, true multicellular organisms and, in combination with oxygenic photosynthesis, giving rise to the ozone belt, it allowed terrestrial life forms (reviewed in Herrmann 1997). The plastids of plants and mitochondria are now known to be extant derivatives of eubacteria, a cyanobacterium and an α-proteobacterium, respectively, incorporated by an as yet unknown heterotrophic host. Both symbiosis events presumably occurred monophyletically and were forced by environmental changes. Molecular work during the past two decades has provided an unprecedented insight into the evolution of eukaryotes, in particular into the evolution of their compartmentalized genome. The contours emerging from these studies suggest that understanding the history of such genomes is a critical piece to the puzzle understanding plant speciation processes.

The subdivided eukaryotic genome has resulted from a massive restructuring and intermixing of the genomes of the initially free-living symbiotic partner cells with loss, intracellular transfer and gain of genetic information. The latter includes lateral gene transfer (summarized in Herrmann 1997; Herrmann & Westhoff 2001). Today, the eukaryotic cell operates with an archebacterial-like transcription machinery and a single, predominantly endosymbiont (eubacterial) type of metabolism (Martin & Schnarrenberger 1997). The majority of organelle genes, including all of those encoding single-chain proteins, were transferred to the nucleus or lost (Herrmann 1997; Martin et al. 1998; Bock & Timmis 2008). The situation is different for supramolecular, multisubunit organelle structures, such as the energetic membrane systems or organelle ribosomes, for which gene transfer was, and probably still is, gradual. As a consequence, plastids and mitochondria still encode part of their energy-transducing respiratory or photosynthetic (thylakoid) membrane systems as well as much of their gene expression machinery. Genome rearrangement had to be accompanied by fundamental changes in expression signals at almost all regulatory levels. Novel networks had to arise to coordinate the intracellular compartmental interaction that did not exist in any of the endosymbiotic partners. These operate in concert with ancient regulatory circuits derived from the ancestral cells (Herrmann 1997; Race et al. 1999; Herrmann & Westhoff 2001). Although the endosymbionts ceded much of the control for their free-living state to the nuclear genome, it is evident that crucial features, including the primary energy sources of the eukaryotic cell or (in plastids) efficient protein synthesis, are regulated not only by the nuclear but by the partite genome.

Most of the genome restructuring occurred at the unicellular level (Martin et al. 1998). With the evolution of multicellular organisms and their functional division of labour in diverse tissues, novel genetic programmes were required, notably for intercellular regulatory networks. In plants, various structural and functional diversifications are known from the plastid (for figures, see Herrmann et al. 1992). Furthermore, in multicellular organisms, growth, division, maintenance, adaptation and modification of organelles or generative processes require not only the coordination of the subdivided genetic machinery and spatially separate protein synthesis machineries but also their integration into diverse ontogenetic programmes. Such programmes evolved gradually. Life cycles and body plans vary significantly from thallophytes (algae, fungi) through mosses to vascular plants (Lopez-Juez & Pyke 2005). A remarkably large fraction of the nuclear-coding capacity, in the order of 25–30%, is required for the management of cell organelles (Herrmann 1997; Martin et al. 2002), and many plastid proteins that regulate organelle biogenesis and function appear not to be of cyanobacterial origin, but encoded by genes that either evolved de novo or were transferred horizontally (Leister 2003; Liere & Börner 2007). Because the basic photosynthetic machinery is highly conserved among oxygenic photoautotrophs (Cramer et al. 2004; Nelson & Yocum 2006), as is the respiratory machinery in general as well, in many cases developmental programmes are probably responsible for the phenotypes observed after interspecific organelle exchanges (Schmitz-Linneweber et al. 2005; Greiner et al. 2008a). This not only emphasizes the crucial role of the entire genetic system in control and metabolic integration of the principal energy sources for cell and organism but also highlights that organellar and nuclear subgenomes constitute a tightly integrated functional unit that co-evolves. Thus, attributing speciation processes solely to the nuclear domain falls short, because it neglects the impact of organellar genetic information on reproductive barriers between populations.

Plastome–genome incompatibility and other kinds of hybrid dysfunctions

The fact that organelle and nuclear genomes must have co-evolved becomes obvious after interspecific or intergeneric organelle exchanges, for instance, of plastids and nuclei. Such exchanges can be obtained by introgression breeding (e.g. Iwanaga et al. 1978; Levin 2003), cybrid technology, i.e. by somatic genetics using cell and tissue culture after protoplast–protoplast or protoplast–(nucleus-free) microplast fusion (e.g. Kushnir et al. 1991; Levin 2003), or using the genetics of permanent-translocation heterozygosity, by which recombination of nuclear genes is largely suppressed (for details, see Rauwolf et al. 2008; Stubbe 1989). An organelle exchange can lead to serious developmental disturbances, based on so-called plastome–genome incompatibilities (Stubbe 1989; Herrmann et al. 2003; Levin 2003). The developmental aberrations affect quite a large number of ontogenetic processes, such as the sexual phase, leaf morphology (summarized in Hagemann 1964; Kirk & Tilney-Bassett 1978) or the photosynthetic and/or respiratory machineries. They can cause hybrid sterility or hybrid inviability (hybrid weakness) (e.g. Stebbins 1950; Stubbe 1989; Levin 2003) and are often recognized as hybrid bleaching or hybrid variegation (Fig. 1). Plastome–genome incompatibility reflects a dysfunctional interaction between the cellular genetic compartments. As opposed to nuclear and plastid mutations affecting the organelle, it is a reversible phenomenon. A dysfunctional foreign plastid will re-green if recombined with its native genome.

Organelle–nuclear incompatibilities and their evolutionary significance are well known from mitochondria (e.g. Lamprecht 1944; Michaelis 1954; Grun 1976; Willett & Burton 2001; Sackton et al. 2003; Fishman & Willis 2006; Lee et al. 2008). Probably, the most prominent example for such a dysfunction in plants is the commercially important cytoplasmatic male sterility (CMS), known from many taxa (Chase 2007). Hybrid nucleo-organelle dysfunctions can result in post-zygotic hybridization barriers and usually show differences in the offspring of reciprocal crosses owing to non-Mendelian inheritance of organelles. Such dysfunctions are frequently observed and designated as asymmetric (Burke & Arnold 2001; Tiffin et al. 2001; Turelli & Moyle 2007; Bolnick et al. 2008). Asymmetry in reproductive isolation appears to be common and taxonomically widespread, especially among plant species. It is important to note that hybrid reciprocal-cross asymmetry is not restricted to organelles. It may also arise from endosymbiotic parasites like Wolbachia, sex-specific suppression of transposable elements, epigenetic effects acting on one gender, transcripts residing in the egg cells, sex chromosomes, and, commonly found in angiosperms, gametophyte–sporophyte interactions or triploid endosperm interactions. (e.g. von Wangenheim 1962; Preer 1971; Grun 1976; Tiffin et al. 2001; Turelli & Moyle 2007).

Hybrid dysfunction owing to species-specific co-evolution between an organelle and nucleus must be based on co-adapted loci, located in two cellular spaces. Products of these loci interact with each other (or a product directly with a locus, e.g. cis-element; Greiner et al. 2008a). These loci evolved along different paths after populations split, and the resulting alleles or products of these alleles may not operate properly anymore in a hybrid individual. In this case, an interspecific exchange of cellular compartments would disturb the interplay of co-adapted components and cause negative ‘cytonuclear’ epistasis (Fig. 2). This model conceptually resembles the Dobzhansky–Muller model that was originally developed to explain hybrid inviability in Drosophila and viewed to reflect nuclear gene interactions (Bateson 1909; Dobzhansky 1937; Muller 1942).

Figure 2.

 Summary of the Dobzhansky–Muller model, modified for nucleo/organelle co-evolution, to explain the generation of reproductive barriers. An ancestral population of nuclear genotype aa (boxed in squares) and the plastid genotype b (boxed in ellipses) reflects co-adapted products of genes a and b interacting functionally. The ancestral population may split into two parts, which are temporally isolated from each other. If a new plastid allele B arises in one of these subpopulations, plant sectors may segregate to homoplastomy and eventually the genetic constitution aa/B will get fixed in individuals derived from such material. In the second subpopulation, a favourable nuclear allele A may arise. Individuals of the genetic constitution Aa/b and aa/b freely mate, and allele A may get fixed in a subpopulation of the genotype AA/b. If the diverged subpopulations mate again, hybrids with the genotype Aa/B will be generated. However, because A and B did not co-evolve, their interaction can be maladaptive. Hybridization then results in reduced fitness of the Aa/B offspring, implying that a post-zygotic hybridization barrier was established. The model shown here reflects a dominant incompatibility (see text).

Nuclear incompatibilities are well documented from hybrids of Drosophila, Xiphophorus and various other taxa including plants (summarized, for instance, in Bomblies & Weigel 2007; Burke & Arnold 2001; Orr 2005; Presgraves 2010; Rieseberg & Blackman 2010). Unlike nucleo-organelle interactions and disregarding interactions involving sex chromosomes, they follow Mendelian inheritance and cause symmetric reproductive isolation in the sense that there is no fitness difference between reciprocal crosses. Because epistasis between nuclear and organelle genomes appears to evolve by comparable genetic mechanisms, the Dobzhansky–Muller concept was recently extended to reciprocal-cross asymmetry (e.g. Burke & Arnold 2001; Tiffin et al. 2001; Turelli & Moyle 2007). For both asymmetric and symmetric hybrid dysfunctions, theoretical models exist for estimating the speed and nature of speciation (e.g. Cruzan & Arnold 1999; Coyne & Orr 2004; Engelstädter & Charlat 2006; Turelli & Moyle 2007; Bolnick et al. 2008; Barton & de Cara 2009; Palmer et al. 2009).

In recent years, substantial progress has been made to uncover the molecular basis of hybrid incompatibilities. However, in contrast to hybrid nuclear and nucleo-mitochondrial dysfunctions, the plastid has not been considered adequately in this line of research, although comprehensive classical literature is available for that organelle (summarized in Kirk & Tilney-Bassett 1978).

Hybridization barriers formed by plastids—the case Oenothera

That plastids can form hybridization barriers and are critically involved in speciation processes became obvious in genetic work on species with a biparental mode of plastid inheritance, notably on the genus Oenothera. Analogous findings have been reported, e.g. from Geranium, Pelargonium, and Medicago (Kirk & Tilney-Bassett 1978). Over the past century, Oenothera was developed as a model for studying plant evolution. It represents a group of flowering plants that grow in well-defined, occasionally endemic habitats with different mating systems (Raven 1988; Levin et al. 2004; Wagner et al. 2007; Evans et al. 2009; Johnson et al. 2010; and citations therein).

The unique position of Oenothera in addressing the role of compartmental co-evolution in speciation is based on both a favourable combination of general genetic features and the existence of a well-developed taxonomy, edaphic ecology, cytogenetics and formal genetics (Cleland 1972; Harte 1994; Dietrich et al. 1997). The former include the possibility of wide interspecific cross-fertility, fertility of hybrid offspring, biparental transmission of plastids and permanent-translocation heterozygotic genomes. Such genomes differ in their chromosome arrangement because of reciprocal translocation of entire chromosome arms, largely eliminate the effects of nuclear recombination and prevent multifactorial segregation during meiosis (Burnham 1962; Cleland 1972; Stubbe 1989; Harte 1994; Levin 2002; Rauwolf et al. 2008). Together, these features allow the exchange of plastids and nuclei as well as the substitution of individual or more chromosome pairs or even of entire haploid chromosome sets (so-called Renner complexes) between species. Organelle exchanges frequently result in developmentally impaired, but generally fertile hybrids, displaying plastome–genome incompatibility. The most comprehensive data set is available from subsection Oenothera (= Euoenothera), the best studied of the five subsections in section Oenothera (Stubbe 1989; Dietrich et al. 1997), which therefore is presented here in some detail to illustrate the impact of plastids in speciation. The findings on Oenothera are of general relevance (cf. also Grun 1976; Kirk & Tilney-Bassett 1978).

In subsection Oenothera, three basic nuclear genomes (A, B and C) occur in homozygous (AA, BB and CC) or stable heterozygous (AB, AC and BC) states and are associated with five basic genetically distinguishable plastome types (I–V). Only 12 of the 30 possible combinations of these genomes and plastomes are green, and of these, only seven exist in nature, in a total of 13 geographically widespread outcrossing and self-pollinating species. The remaining 18 combinations display plastome–genome incompatibility to various degrees. They can be generated artificially or occur naturally as inviable hybrids (Fig. 3). Detailed distribution maps for members of the subsection can be found in Dietrich et al. (1997). The genetics of Oenothera was reviewed by Cleland (1972) and Harte (1994). For a taxonomic revision, see Dietrich et al. (1997). The evolution of the subsection with respect to plastome, population structure and hybridization behaviour was outlined by Stubbe (1964), Cleland (1972), Kirk & Tilney-Bassett (1978), Stubbe & Raven (1979), Stubbe (1989), Dietrich et al. (1997), Stubbe & Steiner (1999). The following paragraphs briefly summarize relevant aspects of this work.

Figure 3.

 Plastome–genome compatibilities/incompatibilities in subsection Oenothera, redrawn from Stubbe (1959, 1989) with permission. A, B and C represent the basic nuclear genotypes, I–V the five genetically distinguishable basic plastomes. Genotypes boxed in red represent naturally occurring species. Minor symbols indicate variances noted for some nuclear subgenotypes.

Subsection Oenothera appears to be monophyletic and to consist of three distinct evolutionary lineages. One lineage with the genetic constitution AA-I is comprised of five species: Oenothera elata, Oe. jamesii, Oe. longissima, Oe. villosa and Oe. wolfii, the first mentioned being presumably the most ancient species of the group (Dietrich et al. 1997). A second lineage with the genetic constitution BB-III consists of the ancient species Oe. grandiflora and its descendant Oe. nutans. The third lineage, CC-V, is represented by the single taxon Oe. argillicola that is endemic to the Appalachian mountains and considered as an early relic of the Oenothera evolution. The three lineages originated in Middle or South America and reached the North American continent in several waves. Recent molecular data suggest that evolution and radiation of the subsection started approximately one million years ago in the middle Pleistocene (Cleland 1972; Levy & Levin 1975; Greiner et al. 2008b).

The presumed ancestral forms of the three lineages, Oe. elata, Oe. grandiflora and Oe. argillicola, are now well separated geographically, and post-zygotic hybridization barriers in form of plastome–genome dysfunctions have evolved. As outlined in Table 1, of all possible crosses between the basic genotypes AA-I, BB-III and CC-V, only a single combination (AB-III) generates viable plants. Because no other pre- or post-zygotic isolation mechanism exists between the ancient species (Cleland 1972; Dietrich et al. 1997), plastome–genome incompatibilities alone are responsible for the isolation of the three basic genotypes from each other—a textbook example for asymmetric interspecific incompatibility of the Dobzhansky–Muller type.

Table 1.   Progenies and phenotypes of crosses between the three homozygous Oenothera lineages (cf. Figs 1 and 3)*
  1. *Note that only one of the possible interspecific offspring of the three lineages, AB-III, is green and viable and all other ones show hybrid inviability because of an incompatible plastome–genome combination.

AA-I × BB-IIIAB-Ilutescent
AA-I × CC-VAC-Idiversivirescent
CC-V × AA-IAC-Vyellow green to yellow
BB-III × CC-VBC-IIIalbina or xantha
CC-V × BB-IIIBC-Vperiodically lutescent

Another well-documented example for the impact of plastids on sexual isolation of species is the mating of Oe. nutans, Oe. parviflora (permanent-translocation heterozygotes) and Oe. argillicola (bivalent-forming homozygote). The natural habitats of these three species overlap, and weak prezygotic hybridization barriers already evolved based on different mating systems and floral morphology. Oe. argillicola is open-pollinating with large flowers, and Oe. nutans and Oe. parviflora are small-flowered and moderately outcrossing, but predominantly self-pollinating (Dietrich et al. 1997). Because the prezygotic hybridization barriers are not very pronounced, natural hybrids between Oe. argillicola (CC-V) and Oe. parviflora (BC-IV) can be observed in the compatible combinations CC-IV and BC-IV, as well as between Oe. nutans (BB-III) and Oe. parviflora (BC-IV) in the compatible offspring BB-III/IV and BC-IV. Therefore, provided plastome–genome combinations permit this, prezygotic hybridization barriers such as self-pollination are not strong enough to prevent gene flow between the species.

Gene flow in the subsection is indeed only prevented if plastome and nuclear genome are incompatible in interspecific F1 hybrids. Such incompatible F1 offspring is produced in the third possible cross between the three species. If Oe. nutans (BB-III) mates with Oe. argillicola (CC-V), no viable hybrids between these species are found. The only possible F1 offspring are the combinations BC-III or BC-V, which both have photosynthetic defects (Table 1 and Fig. 3). Because barriers like self-pollination are leaky for these three species, the plastome serves as the only strong barrier to hybridization. Comparable events occur in the hybridization zone of AA-I species and Oe. biennis (AB-II or BA-III).* Although viable hybrids between these two groups have been described (AA-I, AA-II, AB-II and AB-III), some of the possible hybridizations of this cross, such as AB-I or AA-III, yield incompatible combinations (Fig. 3). Here, the plastome builds an asymmetric hybridization barrier, because viable and inviable offspring can be observed, depending on partners and direction of a cross.

The occurrence of plastome–genome incompatibility in natural populations is underestimated

That plastids can cause reproductive isolation and play an important role in speciation is indisputable for a few examples, but is this phenomenon of general impact in nature? Table 2 provides an overview of the known sexual occurrence of plastome–genome incompatibility. To the best of our knowledge, the list includes all published cases, i.e. from 14 plant genera).

Table 2.   Plant taxa exhibiting plastome–genome incompatibility*
  1. *Only taxa in which hybrid variegation occurs are presented, because only in these instances hybrid bleaching can without doubt be correlated with the plastome (for details, see text). For this reason, Impatiens, in the newer literature commonly referred to as genus showing plastome–genome incompatibility, was excluded. Biparental transmission was not reported for Impatiens (Pandey & Blaydes 1957; Harris & Ingram 1991), and in the commonly quoted reference (Arisumi 1985), chimerical plastome seedlings were not described. Cases of plastome–genome incompatibility produced by cybrid technology or established by introgression breeding are excluded as well, because the influence of other asymmetric effects like from mitochondria is not clear in these instances and the use of this material in terms of identifying speciation barriers is doubtful. Plastome–genome incompatibilities gained by cybrid technology, in turn, are artificial and likely do not reflect primary crossing barriers in nature, although the underlying mechanisms for the phenomena observed may be comparable.

AcaciaFabaceaedecurrens with mearnsiiYellow green to periodically paleMoffett (1965)
CampanulaCampanulaceaeInterspecific hybrids of carpatica involving the variety pelviformisWhite to periodically pale, mostly at the cotyledon stagePellew (1917)
americana interspecific hybrids of different populationsNo further described chlorophyll deficiencyGalloway & Etterson (2005)
GeraniumGeraniaceaebohemicum with bohemicum deprehensumWhite to yellow green, altered flower morphologyDahlgren (1923, 1925)
HypericumHypericaceaeHybrids between acutum, quadrangulum montanum, pulchrum, and hirsutum, as well as further species in not clearly elaborated casesDifferent chlorophyll deficiencies depending on cross and direction; occasionally altered flower colourFarenholtz (1927), Noack (1931, 1934, 1937), Renner (1934), Herbst (1935)
MedicagoFabaceaedzhawakhetica with sativaChlorophyll deficiency, reduced fertilityLesins (1961)
truncatula Jawniel with Mount TaborWhite or pale to periodically paleLilienfeld (1962, 1965)
MenziesiaEricaceaeSee Rhododendron  
OenotheraOnagraceaeInter- and intraspecific hybrids involving most species in the five subsections of the section OenotheraAll kinds of phenotypes described in this article except of altered flowers and influence on pathogen resistanceCleland (1972), Stubbe & Raven (1979), Stubbe (1989), Harte (1994), Dietrich et al. (1997), and others
PassifloraPassifloraceaemenispermifolia with oestediiPale to whiteMráček (2005)
PelargoniumGeraniaceaedenticulatum with filicifolium or radula, interspecific hybrids of citriodorum minor and cordatumWhiteSmith (1915)
zonale Roseum with zonale hort. Stadt Bern, zonale hort. Trautlieb, and inquinansWhite or pale to periodically paleMetzlaff et al. (1982), Pohlheim (1986), Weihe et al. (2009)
Interspecific crosses involving tetraploid material with zonale hort. Juliane, as well as further not clearly elaborated casesYellow white to yellow greenGrieger (2007)
PisumFabaceaesativum subsp. elatius VIR320 with 41 accessions of PisumPale or yellow green to periodically pale; probably plastid transmission is altered and presumably cytoplasmatic male sterility or cytoplasmic female sterility is plastid-dependent; meiotic abnormalitiesBogdanova & Berdnikov (2001), Bogdanova & Kosterin (2006), Bogdanova (2007), Bogdanova et al. (2009), Bogdanova & Galieva (2009)
RhododendronEricaceaeIntergeneric hybrids between Menziesia and Rhododendron and various interspecific and intersectional hybrids in RhododendronDifferent types of chlorophyll deficiency depending on cross and direction,Noguchi (1932), Ureshino et al. (1999), Michishita et al. (2002), Ureshino & Miyajima (2002), Sakai et al. (2004), Kita et al. (2005)
SileneCaryophyllaceaeotites with pseudotitesWhite to yellow or periodically yellow greenNewton (1931)
TrifoliumFabaceaerepens with uniflorum, nigrescens, hybridum, and ambiguumWhite or periodically yellow green; sometimes reduced pollen viabilityPandey (1957), Pandey et al. (1987), Kazimierski & Kazimierski (1970), Przywara et al. (1989), Meredith et al. (1995)
alpestre with heldreichianumGeneral chlorophyll deficiencyQuesenberry & Taylor (1976)
ZantedeschiaAraceaeOne intra- and several interspecific hybrids between four species of the section Aestivae; ordorata with aethiopica in the section Zantedeschia and hybrids between the two sectionsWhite or pale to periodically yellow green; declined pathogen resistance; sometimes plastid transmission alteredNew & Paris (1967), Yao et al. (1994, 1995), Yao & Cohen (2000), Snijder et al. (2004), Brown et al. (2005)

The relatively small number of taxa for which the phenomenon was detected indicates that the topic has not been extensively studied and hence that the influence of plastids in speciation is probably underestimated. This is predominantly attributable to methodical and genetic constraints. The general and most reliable way to detect plastome–genome incompatibility is the observation of hybrid variegation (Fig. 1), which usually results from biparental transmission of plastids and subsequent sorting-out of organelles during cell divisions (Kirk & Tilney-Bassett 1978; Birky 2001). Only in those cases, can a bleached hybrid phenotype reliably be correlated with a particular plastome type, and only this observation allows plastome-based phenotypes to be distinguished from bleached phenotypes caused by other asymmetric determinants, e.g. the mitochondria. Because only approximately one-third of the angiosperm species studied transmit plastids biparentally (Mogensen 1996), the number of taxa in which hybrid variegation can readily be detected is limited.

It is however important to note that uni- and biparental organelle inheritances are not all-or-nothing features, and hence, the two transmission modes often differ only quantitatively (Birky 2001). They can be influenced by genetic constitution and environment. There are various reports for changed, biased or leaky organelle transmission (e.g. Chiu & Sears 1993; Tilney-Basset 1994; Hansen et al. 2007), for instance biparental transmission in Nicotiana and Pisum, which usually display uniparental plastid inheritance (Bogdanova & Kosterin 2006; Ruf et al. 2007). Formal genetics does not allow easy identification of such instances, implying that cases of incompatibility that occur in plants with uniparental plastid inheritance are usually discarded. Furthermore, biparental transmission can be suppressed by incompatibility in Oenothera and Zantedeschia (Chiu & Sears 1993; Yao et al. 1994). Thus, from a genetic point of view, it is plausible that plastome–genome incompatibility and biparental transmission of plastids are separate and not necessarily linked features.

A representative example for difficulties identifying plastome-genome incompatibility is provided by Trifolium. In this genus, hybrid bleaching can frequently be detected, but in only a few instances, the involvement of the plastome was unambiguously demonstrated, although circumstantial evidence suggests that its influence is quite high (see references in Table 2 and quotations therein). Furthermore, plastome–genome dysfunction may not always result in bleached phenotypes. Incompatibilities such as plastid-borne CMS (Stubbe & Steiner 1999), embryo lethality caused by plastids (Stubbe 1963), completed sorting-out in the embryo (Yao et al. 1994) or different photosynthetic performance of two plastid types in green tissue (Iwanaga et al. 1978; Glick & Sears 1994) are probably quite common in plants but remained undetected and may have been disregarded or overlooked. That plastome–genome incompatibility is more frequent than the genetically detected cases becomes also obvious in cybrids, i.e. plants produced somatically in tissue culture, by protoplast/protoplast or protoplast/microplast fusion, from non-crossable species, which carry the nucleus of one species and the plastome of another one. Here, plastid–nuclear incompatibility was frequently observed as summarized by Levin (2003) and Schmitz-Linneweber et al. (2005).

Taken together, plastome–genome incompatibility is probably a general phenomenon (cf. also Grun 1976; Kirk & Tilney-Bassett 1978). As part of compartmental co-evolution, it is an intrinsic characteristic of eukaryotes that can play an important role in interspecific introgression in plants but is underestimated because its detection is often not easy. It is therefore not surprising that systematic work on this topic is rare.

Classification of plastome–genome incompatibilities and relationships with strengths of hybridization barriers

The presumably widespread occurrence of plastome–genome incompatibility in the plant kingdom provided the incentive to look at the genetic nature of these barriers. The substantial body of data and range of incompatible phenotypes, for Oenothera depicted in Fig. 3, suggest that plastome-dependent hybridization barriers can have different causes. Inspection of hybrid bleached phenotypes indicates that these can genetically be classified into at least four genetic categories, each of them with a different impact as a hybridization barrier. The classification that shares elements with a mathematically derived classification of Turelli & Orr (2000) is presumably not confined to genus Oenothera, plants, plastids or the photosynthetic process but applicable to hybrid dysfunctions involving organelles in general.

A first kind of plastome–genome dysfunction, designated dominant type, is found in F1 nuclear hybrids with the plastome of one parent (F1 plastome–genome incompatibility). Most of the examples listed in Table 2 are dominant incompatibilities (see also Figs 1 and 2). In this instance, plastome–genome incompatibility occurs in spite of the presence of a single copy of a compatible genome in heterozygous constitution. In Oenothera, AB-I, AC-I, AC-V, BC-III and BC-V fall into this class (Fig. 3). Dominant plastid–nuclear dysfunctions build strong hybridization barriers, immediately affecting the F1 generation.

A second class is represented by so-called recessive incompatibilities. These differ from the previous class in that the heterozygotes show normal chloroplast development. Plastome III plastids, for instance, which are native to the BB genotype, are also green in combination with AB. Incompatibilities are observed only in the homozygotes, AA-III in this instance, AA-IV, BB-II and CC-II in others (Fig. 3). The evolutionary consequence of recessive dysfunctions is hybrid breakdown. A compatible, bivalent-forming AB-III F1 hybrid segregates 25% AA-III (incompatible), 50% AB-III (green) and 25% BB-III (green) individuals in the F2. This hybridization barrier is generally weaker than that of the dominant type. Recessive incompatibilities have also been observed in Pelargonium and Trifolium as hybrid breakdown of viable F1 generations or F1 backcrosses (Smith 1915; Meredith et al. 1995). Their occurrence is probably underestimated because of the lack of hybrid variegation in subsequent generations. Organelle sorting-out may be completed in all flower organs during the F1 life cycle, and hybrid variegation, therefore, is not transmitted to successive generations (Kirk & Tilney-Bassett 1978; Birky 2001). Because hybrid variegation is the only reliable way to detect plastome–genome disharmony, recessive patterns may frequently be overlooked.

The third category includes co-dominant or additive plastome–genome incompatibilities. Examples in the Oenothera compatibility scheme are AA-V, BB-I, BB-V, CC-I and CC-III (Fig. 3). Such incompatibilities can in principle be caused by two mechanisms. In general, the incompatibility of BB-I, here chosen as an example, suffers from a dominant maladaptive factor between the B genome and plastome I that already appears in an AB-I background. Replacement of the A genome by a further B genome enhances the relatively weak AB-I phenotype, resulting in the strong BB-I incompatibility. Genetically, two explanations are conceivable. First, BB-I can be caused by two linkage groups, a dominant one, already responsible for the AB-I phenotype, plus a recessive one, which becomes obvious in the homozygous BB background. Both linkage groups, solely responsible for dominant or recessive hybrid dysfunctions, respectively, together cause the strong BB-I phenotype. An alternative explanation is inheritance of a single co-dominant linkage group. In this instance, the BB-I phenotype would be caused by a dose effect. A single B genome in the AB-I background displays a weaker phenotype than two incompatible genomes in BB-I. Both cases assume that AB-I and BB-I share an identical factor. Co-dominant incompatibilities can play some role in evolution, because they reinforce an already existing hybridization barrier in the F2. In the chosen example, the incompatible hybrid AB-I, when selfed, produces F2 progeny that are AA-I (green), AB-I (incompatible) and BB-I (strongly incompatible).

The fourth case, chimeric plastome–genome dysfunctions, is characterized by a heterozygous nucleus and a plastome that has an evolutionary history different from both haploid genomes. One example is the combination BC-I. In nature, plastome I is associated exclusively with A genomes, and its combination with BB or CC is disharmonic. Consequently, if BB-I and CC-I are crossed, an incompatible BC-I offspring is not surprising. It is very likely that such plastome–genome dysfunctions are of polygenic origin. Other incompatibilities of that kind in the compatibility chart are AB-V and, in some respect, AC-III and BC-II) (Fig. 3). The significance of chimeric hybrid dysfunctions to generate hybridization barriers is limited. Their occurrence in nature is improbable, because both parental lines have to be incompatible already and crosses between those parents are usually difficult to achieve.

The functional consequences of plastome-based hybrid dysfunctions in Oenothera can readily be recognized in the compatibility chart (Fig. 3). The degree of lesions is consistent with tree or divergence time calculations (Greiner et al. 2008b) and appears to reflect differences in the evolutionary diversification of the five basic plastomes. For instance, the probably most recently evolved genotype C of the subsection exerts a dominant negative effect on plastomes associated with A and B genomes. On the other hand, plastome I, considered to be the most advanced plastome, is compatible exclusively with its natural AA background. A comparable case was reported from Rhododendron (Noguchi 1932). This suggests that the strength of post-zygotic barriers owing to compartmental incompatibility is related to the degree of ‘cytoplasmic divergence’ (Levin 2003).

Selection pressure acting on the plastome

As outlined above, plastome–genome disharmony can lead to hybridization barriers of different strengths, but which selection forces produce them? That strong forces can act on the plastome is evident from amino acid substitution rates among the five basic evening primrose plastomes which uncovered remarkably high means of ratios of nonsynonymous to synonymous substitutions (Ka/Ks) (Greiner et al. 2008a). They indicate substantial selection pressure on these plastomes, in relatively short timescales (≤Myr). Similar findings were reported for clpP in Silene and for Geraniaceae plastomes (Erixon & Oxelman 2008; Guisinger et al. 2008). Both taxa also display plastome–genome incompatibility (Table 2). The consequence for selection on a distinct plastid type is chloroplast capture, the introgression of a plastid of one species into another. This can occur relatively fast, even if only a small fitness advantage exists, usually for the plastid of the maternal parent (Tsitrone et al. 2003). It was observed within the Triticean tribe, for instance (Redinbaugh et al. 2000).

Selection can be influenced by environmental factors, mutation pressure or selfish genome evolution. Ecological selection favouring a particular cytoplasm has been described from various taxa (e.g. Case & Willis 2008; Sambatti et al. 2008; and citations therein). The photosynthetic machinery and its adaption to changing environment is an obvious target. Higher photosynthesis rates are usually associated with growth and fitness advantage (Arntz & Delph 2001). Differences in photosynthetic performance have been noted with different cytoplasms in introgression lines of Triticum and Aegilops (Iwanaga et al. 1978). Similar findings are known from evening primroses (Glick & Sears 1994) and Ipomopsis (Wu & Campbell 2007). Changes in adaptive impact in ribulose bisphosphate carboxylase/oxygenase (RuBisCO), the key enzyme of the photosynthetic dark reaction, were reported from Schiedea (Kapralov & Filatov 2006). Photosynthesis is strongly influenced by processes such as water balance, light, temperature or oxidative stress, which suggests that periods of climate changes may affect selection for or against minor variation. For instance, evening primrose species grow in quite different habitats (Dietrich et al. 1997). Because their differentiation correlated with a fluctuating climate for both precipitation and temperature during the Pleistocene (Cleland 1972; Evans et al. 2009), it is conceivable that these conditions applied selection pressure.

Competition between two different plastid types expressed by differences in their multiplication rates, a second example for illustrating selection, is well documented for plants displaying biparental transmission such as Oenothera (reviewed in Chiu et al. 1988; Harte 1994) and Pelargonium (Hagemann 1976; Abdel-Wahab & Tilney-Bassett 1981; Weihe et al. 2009). In sexual crosses including two compatible plastids, only the faster multiplying type is found in nature (cf. plastome I > II in Fig. 3). Such differences are heritable and largely independent of compatibility relations. The findings also illustrate constraints owing to compartmental interdependence that are characteristic of eukaryotic genome evolution; a ‘progressive’ mutation in any one of the compartments, as presumably enhanced multiplication rates of plastids, has to satisfy the requirement that the entire system remains fit and genetically balanced (Grun 1976; Herrmann & Possingham 1980). Substitution would not be of benefit if the latter is incompatible. It will supplant the slower multiplying, compatible organelle, and the plant may die (cf. also Grun 1976). Such multiplication rate differences may also influence reproductive success, e.g. levels of asymmetry. Seed and pollen parent decide about the magnitude of segregation, and implicitly gene flow, owing to different numbers and physiological status of plastids between egg and generative cells (Meyer & Stubbe 1974). A maternal parent with a faster but incompatible plastid would not be advantageous in hybrid crosses. However, used in reverse direction, the retarded development of the pollen plastids would mitigate deleterious effects of the faster multiplying organelle and not necessarily eliminate the slower multiplying, compatible plastid type. It will be interesting to determine whether the underlying molecular principle towards changes in plastid multiplication rates resides in or acts on the DNA molecule itself, e.g. during replication (Hornung et al. 1996; Sears et al. 1996).

It is important to note that the genetic determinants for multiplication rate differences are predominantly plastome-encoded traits (reviewed in Chiu et al. 1988). This is different from developmental adjustments of plastome-to-genome ratios (Rauwolf et al. 2010). Cellular plastome dosages are likely under efficient nuclear control, because competitive segregation of two organelles occurs within a given genome-to-plastome ratio according to comparable chloroplast sizes and numbers per cell. No ‘conflict situation’ in replication or multiplication rates between nuclear and organelle genomes has yet been reported that would significantly influence the intracellular balance of the cellular subgenomes.

It has been proposed that uniparental transmission of organelles evolved to avoid cytonuclear conflicts. In contrast to biparental inheritance, it should ensure co-evolution between nuclear and organelle genomes and prevent competition between native and foreign cytoplasms, e.g. by differences in organelle multiplication rates (Grun 1976; Cosmides & Tooby 1981; Hoekstra 2000). On the other hand, uniparental organelle inheritance excludes recombination between organelle genomes of different mating types which could spare the asexual organelle from Muller’s ratchet. Homologous recombination of organelle genomes may therefore act as a driving force for evolution and maintenance of biparental transmission (Grun 1976; Birky 1995; Hoekstra 2000; Barr et al. 2005). However, in sexual hybrids of vascular plants, homologous recombination of plastid genomes is entirely suppressed (Chiu & Sears 1985). Hence, it is an unresolved paradox that biparental transmission along with a lack of recombination in plastids exists in so many plant taxa. The issue is controversial and remains to be settled (e.g. Sears 1980).

Physiology and cell biology of hybrid plastome–genome dysfunction

Abundant effects of plastome–genome dysfunctions concern various kinds of albinism. Such phenotypes may display relatively fast necrosis, even before or during germination. Other alterations are ontogenetically less severe, and plants may even fully re-green temporarily or later in development (Table 2 and Fig. 3). Generally, incompatible hybrid materials suffer from reduced pigment content, lower rates of photosynthesis and an impaired thylakoid structure. If multisubunit photosynthetic membrane assemblies do not operate correctly, they are often unstable, because scavenging systems that operate with high subtlety remove or reduce them (e.g. Herrmann 1996; Amann et al. 2004). Occasionally, plastids in leaf tissue degenerate completely, or organelle ribosomes are lost. Morphological and physiological analyses of bleached phenotypes that are available from Zantedeschia (Yao & Cohen 2000), Passiflora (Mráček 2005) and Oenothera (Johnson & Sears 1990; Glick & Sears 1994; Harte 1994; Dauborn & Brüggemann 1996; Greiner et al. 2008a) suggest that such changes can occur at many regulatory or structural levels.

A promising approach to identifying molecular components involved in plastome–genome incompatibility rests on an analysis of retrograde (plastid) and anterograde (nuclear) signalling (Woodson & Chory 2008). The Oenothera plastome–genome combination AA-III (Fig. 3), chosen as an example, bleaches reversibly, and circumstantial evidence suggests that this is attributable to a temporally dysfunctional differentiation process of the chloroplast. It may partially be neutralized by growth hormones (Glick & Sears 1994) and can be cured genetically if plants carry the compatible plastome II in tissues adjacent to that of the incompatible plastome III. In mixed cells, in which sorting-out of the two plastome types is not complete, and in cell layers containing plastome III oriented ad- or abaxially to at least one cell layer with plastome II, the incompatible plastome is able to re-green in an AA background (Stubbe 1958); for, pictures see p. 167, plate 2e, in Harte (1994). Here, the missing component seems to be a metabolite or gene product, which can cross cell borders and move from organelle to organelle.

Another spectrum of quite abundant hybrid nucleo-organellar phenotypes suffers from general cellular dysfunction or/and an impaired sexual phase. Inhibition of cell growth can cause embryo lethality, lack of germination and changes in leaf morphogenesis or in other organs (Stubbe 1963). Such developmental disturbances may reflect only pleiotropic effects owing to malfunctioning of plastids or/and mitochondria, because the organelles are crucial components of the metabolic and signalling network of the cell (Laloi et al. 2006; Pesaresi et al. 2007). Examples are known from some CMS phenotypes affecting flower morphology which possibly result from reduced mitochondrial ATP levels that cause misexpression of floral regulators (Chase 2007). Severe effects on cell growth, secondarily affecting leaf or flower morphology, are known from a general inhibition of plastid translation (Ahlert et al. 2003) or from various plastome knock-out lines, such as accD (Kode et al. 2005), clpP (Shikanai et al. 2001), rps18 (Rogalski et al. 2006), ycf1 and ycf2, two ORFs of unknown function (Drescher et al. 2000).

The haploid ontogenetic phase, the gametophyte, can be affected as well causing cytoplasmic female sterility (CFS), CMS or reduced pollen vigour (Stubbe et al. 1978). Both gametophytic and sporophytic effects can also have a strong impact on biparental or uniparental transmission of plastids. In extreme cases, transmission of a strongly incompatible plastid can be suppressed completely (Chiu & Sears 1993; Yao et al. 1994). Furthermore, male sterile anthers are frequently associated with round, rather than spindle-shaped, starch grains in the pollen (Stubbe & Steiner 1999). Such pollen grains usually exhibit altered respiration, lipid and starch metabolism (Göpel 1976). In some instances, plastome-dependent pollen abortion is correlated with chromosome breaks, asymmetric anaphase chromosome distributions and trinucleated tetrads (Chapman & Mulcay 1997). In Hypericum, flower colour may depend on plastome–genome interaction (Farenholtz 1927), as well as pathogen defence in Zantedeschia (Snijder et al. 2004). For literature on Oenothera, see Harte (1994); further work is cited in Table 2 and quoted above.

Plastome–genome dysfunction appears primarily as a regulatory phenomenon—molecular analysis

An increasing body of information suggests that RNA modification or metabolism is one of the key players in plastome–genome hybrid incompatibility. When well-conserved operons between cyanobacteria and plastids are compared, RNA metabolism in higher plants is substantially changed. In contrast to cyanobacteria, primary transcripts from plastids are extensively processed in the organelle, presumably to account for the more complex situation of eukaryotic genome expression (Herrmann & Westhoff 2001; Liere & Börner 2007; and citations therein). Several kinds of RNA processing are rapidly evolving. For instance, the plastomes of Arabidopsis thaliana ecotypes can differ in their editotypes (Tillich et al. 2005), and intraspecific variation of 5′ and 3′ ends was found for mitochondrial mRNA from that plant (Forner et al. 2007). In two cases with photosynthesis defects, the plastid determinants of hybrid dysgenesis are known. They are involved in different regulatory levels. Failure to edit just a single nucleotide in atpA, a gene encoding the α-subunit of the thylakoid located ATP synthase, is responsible for bleaching of the cybrid with the Atropa genome and the Nicotiana plastome. Furthermore, species-specific RNA editotype differences also explain the enormous phenotypic difference of the reciprocal Nicotiana/Atropa cybrid. The former one is nearly white, while the latter combination is fully green (Schmitz-Linneweber et al. 2001, 2005). A regulatory sequence interval of Oenothera plastome I, in turn, the intergenic promoter-bearing region between the divergently transcribed clpP and psbB operons, appears to be the cause of compartmental incompatibility in the evening primrose AB-I hybrid (Greiner et al. 2008a). The corresponding nuclear partner genes, in the latter case presumably transcription factors, remain to be identified. Comparable cases were reported for nucleo-mitochondrial incompatibilities. From several plant species, the mitochondrial determinants of CMS are known: short toxic polypeptides decoded from novel rearranged chimeric ORFs and products of nuclear loci restoring fertility by interaction with the mRNAs of these polypeptides (Linke & Börner 2005; Chase 2007).

The findings are consistent with those of other incompatible hybrid cases, in which gene regulation plays an important role as well (Haerty & Singh 2006; Ortiz-Barrientos et al. 2007). For instance, incompatible Zantedeschia (Yao & Cohen 2000) or Passiflora plastome-genome hybrids (Fig. 1, Mráček 2005) suffer from altered plastid gene regulation. In Rhododendron, changes in nuclear genome dosage were probably responsible for overcoming plastome–genome dysfunction (Sakai et al. 2004). Altered gene and metabolic regulation rather than modified basic cellular structures appears to be a driving force in speciation of Zea mexicana (Teosinte) and Zea mays (Doebley 2004). Diverse molecular functions of genes encoding interacting partners of the Dobzhansky–Muller type were identified from nuclear or nucleo-mitochondrial hybrid inviability or breakdown lines from Mus, Xiphophorus, Drosophila, Tigriopus, Saccharomyces, Arabidposis, Mimulus and Oryza. They include genes for transcription and translation factors, chromatin or RNA-binding proteins, a receptor tyrosine kinase, an acylphosphatase, or components of the nuclear pore (Orr 2005; Michalak & Ma 2008; Barr & Fishman 2010; Presgraves 2010; Yamagata et al. 2010), substantiating that hybrid dysfunctions are often of regulatory nature and can occur at various levels. For a review of Dobzhansky–Muller incompatible plants, see Rieseberg & Blackman (2010).

In principle, hybrid incompatibility could reflect a regulatory or/and a structural phenomenon (Hoekstra & Coyne 2007). Speciation-relevant direct alteration of structures, such as the photosynthetic machinery or the plastid ribosome, could play a role in plastome-genome incompatibility, because these structures frequently display dysfunctions. However, so far, no evidence in support of this view has been obtained. Multisubunit organelle structures are highly conserved, e.g. the oxygenic photosynthetic machinery from cyanobacteria to vascular plants (Cramer et al. 2004; Nelson & Yocum 2006). The basic membrane system was streamlined and optimized quite early—billions of years ago—and hence is not expected to play a principal role in microevolution any more. This is in line with observations that variant structural thylakoid proteins of plastid or nuclear origin, such as psbB, a chlorophyll a binding protein of the photosystem II core complex, or atpA, are probably not relevant for incompatibility in evening primroses (Greiner et al. 2008b). On the other hand, the ontogenetic programmes into which the biogenesis of those machineries must be integrated change substantially during evolution and are under selection. The finding that not the structure of the photosynthetic machinery itself, but primarily regulatory events during its biogenesis and function cause compartmental incompatibility could therefore indicate an important principle in speciation, e.g. adapting or fine-tuning of organelle and photosynthetic process to distinct situations and/or habitats. This is not unexpected, considering that gene expression and biogenetic processes require about one order of magnitude more genes than the structural components alone (Herrmann & Westhoff 2001). Presumably, adaptation and speciation rest on a combination of regulatory and structural mutations, with more impact of the former. Because the molecular biology of the photosynthetic machinery and organelle is well understood, it will be a fascinating task to examine this hypothesis in more detail.

Nuclear genetics of plastome-genome incompatibility

At present, it is difficult to estimate kind and number of nuclear or plastid loci involved in hybrid dysfunction in distinct incompatible combinations or at all. Molecular data are scarce, and incompatible F1 hybrids are sometimes sterile or exhibit reduced fertility (Table 2), complicating genetic analysis of determinants. For Pelargonium (Smith 1915), Pisum (Bogdanova & Berdnikov 2001), Medicago (Lilienfeld 1965) and the Oenothera BB-II combination (Stubbe 1953), monogenic nuclear determinants were implied by genetic analysis. The only two above-mentioned cases, from which the plastid determinants for plastome–genome dysfunctions are known, are monogenic as well. However, gene-to-gene interactions of the Dobzhansky–Muller type may reflect only a part of more general and more complex evolutionary scenarios. Segregation analysis in Zantedeschia, for example, uncovered two or three nuclear loci that cause virescent phenotypes (New & Paris 1967; Yao & Cohen 2000—for pictures of Oenothera see Fig. 1). In Pisum, two unlinked dominant loci, designated Scs1 and Scs2, contribute to plastome-genome incompatibility (Bogdanova et al. 2009). Two or more nuclear loci also appear to be involved in various incompatible hybrids in Oenothera (van der Meer 1974; Jean 1984; Rauwolf 2008). In such a case, the marker lor combined with an incompatible plastome results in embryo lethality (Renner 1943), but only if linked to marker Fl. Only lor lor Fl Fl genotypes lead to embryo lethality. Individuals homozygous for lor alone combined with a foreign plastome suffer only from chlorophyll deficiency. These examples illustrate also that nuclear loci involved in compartmental incompatibility can be genetically dissected, and are thus amenable to molecular analysis.

Experimental perspectives and models

An important benefit of working with plastome-genome dysfunctions rests in substantial knowledge about the photosynthetic machinery and chloroplast genomes, which only house a limited number of genes (in order of 120; Sugiura 1989). Genetically defined lines for plastome–genome incompatibility promise relatively easy access to speciation-relevant molecular determinants, their physiological function and possibly ecological consequence compared to other models currently under study. The quite diverse phenotypes suggest that a relatively broad spectrum of genes or processes may become amenable to molecular analysis. They usually exert little direct effect on fertility and often have more subtle effects on hybrid viability than hybrid sterility that is predominantly being studied so far. Materials, not obtained by sexual crossings, could be instrumental to study distinct questions, for instance Solanacean species from which controlled tissue culture and somatic genetics, notably cybrid technologies (Kushnir et al. 1991) and plastid transformation (Schmitz-Linneweber et al. 2001, 2005) are available. However, the prospects are generally limited and usually secondary, not causative events for speciation will presumably be detected. The value of this material to elucidate primary speciation events is therefore questionable. This illustrates a basic problem with working on a functional genetics of speciation, namely the difficulty distinguishing, whether the so-called speciation genes are the initial course of a reproductive barrier or just a by-product of speciation, only strengthening an already existing hybridization barrier (cf. Coyne 1992). The latter is especially relevant for reproductively isolated, related organisms that do not exchange genes.

Choices for molecular genetic work on plastome–genome incompatibility are hence limited to species listed in Table 2. Unfortunately, for most materials, including Medicago and Pisum, the genetic basis for investigating compartmental incompatibility is meagre and often based on single crosses (Table 2). Thus, classical genetics, phylogeography as well as comparative molecular analysis of these materials would need to be developed first. At present, various kinds of incompatible phenotypes have been reported only from Rhododendron, Hypericum, Trifolium, Zantedeschia and Oenothera (Table 2), and strictly speaking, at first glance, only these genera would allow more detailed access to underlying speciation forces. For obvious reasons, Rhododendron would not be an appropriate choice, and many hybrids of Hypericum, Trifolium and Zantedeschia suffer from hybridization barriers, such as uneven chromosome numbers, hybrid sterility, embryo abortion or other limitations preventing success of interspecific crosses (see references in Table 2). Oenothera lacks these serious genetic limitations and hence is an unrivalled material for such studies. In evening primroses, genetically different plastome types are identified and their distribution in various species and impact on evolution have been investigated, including an intensive phenotypic, genetic, cytogenetic and physiological characterization of plastome–genome incompatibility (Cleland 1972; Harte 1994; Dietrich et al. 1997).

In evening primroses, morphologically different and interbreedable species together with biparental transmission of plastids and a general interfertility of species including interspecific plastome–genome offspring are the rule. If sterile offspring occurs in exceptional cases, sterility is plastome-dependent (CMS or CFS) and can be cured genetically by supporting an incompatible, sterile combination with an additional, compatible plastome type (Stubbe et al. 1978). Other hybridization barriers do not play a notable role in speciation, at least in subsection Oenothera, on which most of this type of work was performed (Dietrich et al. 1997). However, the approach is not limited to that subsection, although the other subsections have not yet received a comparably intense genetic study. Quite substantial information regarding interfertility and plastid/nuclear compatibility relationships for subsections Raimannia, Munzia, Nutantigemma and Emersonia is available. Their hybrid offspring are usually fertile. Even the generation of intersubsectional compartmental hybrids is possible, these are often extremely disharmonic and their fertility is sometimes reduced (Stubbe & Raven 1979). On the other hand, intraplastome variation in DNA restriction patterns, which was noted for all five Oenothera plastomes (Herrmann et al. 1980; Rauwolf et al. 2008), is genetically virtually neutral. Whether such differences are correlated with subpopulations of species with floating chromosome arrangements that grow in well-separated habitats over an entire continent (Cleland 1972) and reflect phylogenetically relevant prespeciation processes, and/or correlate with geographical or environmental trends remains to be seen. Monitoring mild dysfunctions in hybrids from defined diverging populations that are usually at branch points of speciation may reveal processes and loci involved early in divergence and aid to uncover primary, causative effects of speciation rather than secondary events that only strengthen an already existing barrier. This is especially the case if plastome–genome incompatibility builds a major hybridization barrier.

Concluding remarks

Hybrid nucleo-plastid incompatibility is an important barrier for interspecific gene exchange and post-zygotic reproductive isolation. It is presumably widespread and appears to occur in a similar fashion as hybrid nuclear incompatibility predicted by the Dobzhansky–Muller model. It shares negative epistasis between interacting alleles as the causative principle with that concept, but, as a consequence of the subdivided integrated eukaryotic genome, differs in that it involves co-evolution of nuclear and organelle loci. These usually cause asymmetric hybridization barriers between divergent populations. The significant role of the plastome in speciation is evident from its influence of quite a wide range of ontogenetic traits. Because of its limited and defined coding potential, the plastome (and the well-understood photosynthesis process) should therefore allow relatively easy general and causal access to phylogenetic questions and hypotheses and aid to evaluate the generality of the negative epistasis concept. Identifying loci causing quite diverse hybrid nucleo-organelle dysfunctions is therefore of especial interest that could shed light on molecular aspects, impact, potential costs and selection pressures on plastids during speciation, notably on their co-evolution with nuclear genomes. Based on molecular work on disharmonic plastome–genome hybrids, co-evolved interacting regulatory elements emerged as a basic principle in speciation, at least of compartmental dysgenesis. Such hybrids or cybrids may be considered as ‘network mutants’. Their study could complement mutant approaches or high-throughput analyses in genomics and help to uncover processes involved as well as principles, how such regulatory circuits and networks are designed, the ways they operate and how they change during speciation.


Asymmetric hybridization barriersBreeding barrier acting on only one gender.
CybridA hybrid between the nucleus of one species and the plastome of another. The term is generally used for hybrids made by protoplast/protoplast or protoplast/microplast fusion, which therefore share initially the organelle genomes of both fusion partners.
EditotypeSpecies- or line-specific pattern of all editing sites in the transcripts of an organelle genome.
EpistasisModification of the function of a gene by another one.
Gametophytic lethal factorA factor or locus aborting the haploid generation, i.e. egg cell or pollen.
Hybrid bleachingA form of hybrid inviability caused by chlorophyll deficiency.
Hybrid breakdownOccurrence of maladaptive traits in higher generations (F2 or F3 breakdown) owing to the disturbance of a co-evolved and genetically linked gene cluster, a so-called supergene.
Hybrid inviabilityWeakness or inviability of a hybrid.
Hybrid variegationA hybrid that bears two parental (wild-type) plastids, of which only one is compatible with the nuclear genome, while the other is not. Both organelle types may segregate into chimeras during plant development. The sorting-out of different plastid types and their co-existence within individual cells demonstrate that the observed differences reside in the organelles themselves, that they are genetic and that an impaired plastid cannot be simply cured by an exchange of metabolites.
Introgression breedingBreeding technology which allows a genetic substitution of cytoplasms. Using this technique, the maternal parent is backcrossed for several generations with the pollen donor. After several rounds of backcrossing, the maternal nuclear genome is replaced by the paternal one and now harbours the maternal organelles.
MaladaptiveA trait that reduces the fitness of an individual.
PlastomePlastid or chloroplast genome.
Permanent-translocation heterozygosity (syn. structural heterozygosity, complex heterozygosity or terminal translocation heterozygosity)Inheritance of entire parental haploid chromosome sets (Renner complexes) as units, resulting in identical progeny owing to suppression of chromosomal recombination and segregation and the existence of a system of lethal factors or self-incompatibility.
Prezygotic and post-zygotic hybridization barriersBreeding barrier established before and after fertilization, respectively.
Sexual, generative or reproductive phaseDevelopment of sexual processes and organs in a life cycle. Higher plants possess a so-called heterophasic or intermediary generation cycle (different from the diplontic generation cycle of higher animals), with flower organs gametophyte, gametes and sporophyte (seedling formation after fertilization and individual).
Sorting-outVegetative segregation of two genetically distinct plastids types in the course of cell division during ontogenesis. Starting from ‘mixed cells’, harbouring two genetically different plastid types, plastids are distributed randomly to the daughter cells. As a stochastic process, this regularly leads to two cell lineages, containing only one of the two plastid types, respectively. These cell lineages give rise to variegated plant tissue.
Sporophytic lethal factorA factor or locus aborting an early diploid generation, generally pre-embryo or embryo stages.
Water-use efficiencyRatio of photosynthetic to transpiration rate.


  • *

    Oe. biennis is a permanent-translocation heterozygotic species with the basic Renner complexes A and B each of which is typically linked to one sex. BA-III corresponds to subgroup biennis-1 with B as maternal (egg cell) and A as paternal (pollen) complex, whereas AB-II describes subgroup biennis-2 with reverse sexual linkage of the haplo-complexes. For details, see Cleland (1972).

  • Circumstantial evidence based on formal genetic data suggests that incompatibility in Epilobium does not affect interactions of plastids with the nuclear genome alone but also exists between plastids and mitochondria (Michaelis 1954).

  • The combinations AC-III and BC-II are chimerical incompatibles only if they originate from crosses between their incompatible parents AA-III × CC-III and BB-II × CC-II, respectively. As an exception, the genetics of permanent-translocation heterozygosity in Oenothera allows the assembly of AC-III or BC-II from compatible parents (e.g. AB-III × CC-V or BA-II × CC-V). In this case, AC-III or BC-II represent dominant incompatibles.


We thank Barbara B. Sears, Rudolf Hagemann, Christian Schmitz-Linneweber, Wolfgang Stephan, Diethard Tautz and three anonymous referees for helpful comments and discussions, Dr. Elizabeth Schroeder-Reiter for language editing, and Josef Bergstein for photographic service. This research was supported by grants of the Deutsche Forschungsgemeinschaft (SFB TR1 and He 693/16 to R.G.H and J.M.), the Hanns-Seidel-Stiftung supported by the Bundesministerium für Bildung und Forschung (S.G.), and the Max-Planck-Society (S.G.).

S.G. is working on cytoplasmic speciation barriers, using primarily the model plant Oenothera. His work includes mechanistic and evolutionary aspects of the suppression of homologous recombination in functional apomictic plant species. U.R. is interested in the model Oenothera as well, in new views of the diversification of life, in new genetic system for medical application, and further in the development of new substance for the pharmaceutical industry. J.M. is molecular geneticist, whose research concentrates on various aspects of chloroplast biology, including photosynthesis, RNA metabolism, biogenesis, evolution and signaling. R.G.H. research focuses predominantly on organelle biology, notably on plastids, their genomes including their role in the biogenesis and function of the photosynthetic machinery, and in the evolution of the eukaryotic domain.