Review Article

The evo-devo of plant speciation

  • Nature Ecology & Evolution 1, Article number: 0110 (2017)
  • doi:10.1038/s41559-017-0110
  • Download Citation
Published online:


Speciation research bridges the realms of macro- and microevolution. Evolutionary developmental biology (evo-devo) has classically dealt with macroevolutionary questions through a comparative approach to distantly related organisms, but the field later broadened in focus to address recent speciation and microevolution. Here we review available evidence of the power of evo-devo approaches to understand speciation in plants at multiple scales. At a macroevolutionary scale, evidence is accumulating for evolutionary developmental mechanisms giving rise to key innovations promoting speciation. At the macro microevolution transition, we review instances of evo-devo change underlying both the origin of reproductive barriers and phenotypic changes distinguishing closely related species. At the microevolutionary scale, the study of developmental variation within species provides insight into the processes that generate the raw material for evolution and speciation. We conclude by advocating a strong interaction between developmental biology and evolutionary biology at multiple scales to gain a deeper understanding of plant speciation.

Speciation research addresses the evolutionary processes that generate the extraordinary diversity of life on Earth as well as the patterns derived from them1. It bridges the realms of macro- and microevolution, respectively dealing with evolutionary phenomena above and below the species level2. This transition is defined by the establishment of reproductive barriers restricting gene flow between populations, the prerequisite for species formation under the biological species concept. Speciation studies have classically focused on genetic and ecological mechanisms, essentially ignoring the role of developmental mechanisms. At the same time, the more recent field of evolutionary developmental biology (evo-devo), which aims to understand the developmental mechanisms of evolutionary change3, has usually dealt with macroevolutionary questions through a comparative approach with distantly related organisms4, paying less attention to speciation and microevolutionary processes. A broadening of focus in evo-devo to address recent speciation and microevolution has been advocated5,6. Indeed, new research shows the potential of a multidisciplinary approach including evo-devo to provide deeper insight into speciation.

Plants, and particularly the outstandingly diverse angiosperms, provide excellent opportunities for this approach. A number of model systems for the study of speciation have been characterized genetically, developmentally and ecologically. Moreover, several genes involved in developmental processes potentially generating reproductive isolation have been characterized7,8. Here we review available evidence for the power of evo-devo approaches to understand speciation processes and patterns in plants at multiple scales. Our focus is mainly on evolutionary changes in developmental patterns, or ‘developmental repatterning’3, underlying reproductive barriers and other phenotypic differences arising during or shortly after speciation.

Although it is difficult to quantify the role of developmental changes in reproductive isolation relative to other mechanisms, available data suggest that developmental repatterning is particularly relevant to pre-zygotic barriers, and specifically to those acting at the pre-pollination level. In a review of speciation genes in plants8, all examples of speciation genes underlying pre-pollination barriers were involved in developmental processes. These included genes causing temporal isolation and pollinator isolation, which are frequent targets of speciation studies in flowering plants (although it has also been argued that pollinator specialization can frequently be a consequence, and not a cause, of speciation9). Post-zygotic barriers, on the other hand, include mechanisms of hybrid inviability and hybrid sterility7,8 that are not described as developmental repatterning, and therefore are not reviewed here. Pre-zygotic barriers are thought to contribute more than post-zygotic barriers to reproductive isolation in plants7, which indicates that developmental repatterning plays a crucial role in plant speciation.

Macroevolutionary scale

Evo-devo research has frequently focused on comparing developmental processes across species separated by large evolutionary distances. These research lines fall within the realm of macroevolution, which is generally regarded as the long-term cumulative result of microevolutionary mechanisms10. However, some macroevolutionary patterns may not be entirely explained in this way, such as the differential diversification of clades (species selection) and the ways in which such differential diversification is related to morphological variety (disparity)11,12.

Speciation-promoting traits. Bursts of speciation (radiation) may be the result of the evolutionary acquisition of ‘key innovations’: an evolutionary change in a trait that is causally linked to an increased diversification rate in the resulting clade13. This may be the result of exposing the lineage to new areas of phenotypic space and new ecological opportunities12,14. Evo-devo can provide information on the microevolutionary developmental mechanisms by which key innovations first evolved, along with insights into their macroevolutionary effect on speciation patterns.

The evolution of many species-rich plant clades was preceded by whole-genome duplication (WGD, or polyploidy) events that allowed the diversification of regulatory genes involved in the development of key innovations15,16. Frequently, however, there is a lag between innovation and radiation17. It has been suggested that, after an innovation first appears, strong developmental robustness needs to evolve for natural selection to efficiently explore phenotypic and ecological space, thus leading to both species diversification and an increase in morphological disparity12. As an outstanding example, the angiosperm flower can be considered a combination of key innovations that fostered diversification through the exploration of a brand-new array of plant–pollinator interactions18. Although the developmental origin of the angiosperm flower has not yet been explained, it is known that a WGD event occurred before the diversification of angiosperms, and duplication of homeotic MADS-box genes controlling the identity of floral organs was likely important in the origin of the flower15,16. A role for developmental robustness in the subsequent radiation is supported by the observation that early-diverging lineages of angiosperms are relatively species-poor and display a low degree of floral developmental robustness, in contrast with the high robustness found in some of the most diverse angiosperm families, such as the Orchidaceae12.

Aside from the actual flower itself, additional floral key innovations independently acquired in multiple families are thought to have consistently enhanced diversification within the angiosperms14. The developmental mechanisms recruited in these independent origins of key traits may or may not be the same between species at the organ level (for example, different organ identity, contrasting growth patterns), cell level (for example, patterns of cell division and cell expansion) or molecular level (for example, changes in regulatory or coding regions of the same or different genes). Thus, recurrently acquired key innovations constitute phylogenetic replicates that can be used as ‘metamodels’ in which to test the consistency of those developmental innovations promoting speciation across lineages19. For example, floral zygomorphy (bilateral symmetry) is known to have evolved many times from actinomorphic (radially symmetric) ancestors, and is thought to have promoted speciation by enabling specialized animal pollination20. In eudicots, the evolution of zygomorphy has recurrently involved the recruitment of CYCLOIDEA-like genes that are dorsally expressed during flower development21. In contrast to zygomorphy, diverse molecular mechanisms might underlie the repeated evolution of nectar spurs, another floral key innovation (Box 1).

Box 1: Nectar spurs and speciation.

Floral nectar spurs constitute one of the best examples of a trait involved in plant speciation that is being studied at multiple evolutionary scales, integrating evolutionary, developmental and ecological perspectives. These spurs are tubular outgrowths of floral organs usually containing a nectar reward for pollinators, and they have evolved multiple times during the diversification of flowering plants, within families as distantly related as Ranunculaceae, Orchidaceae, Violaceae and Plantaginaceae87. Phylogenetic comparative analyses suggest bursts of diversification associated with clades with nectar spurs, leading to the consideration of this trait as a ‘key innovation’ promoting speciation through pollinator specialization14,87.

Macroevolutionary scale. The comparative study of nectar spurs in different families allows us to test the degree to which similar ontogenetic and genetic mechanisms have been recurrently recruited to produce a morphologically and functionally convergent trait. Thus far, ontogenetic mechanisms producing nectar spurs seem to be broadly similar across families, with an initial phase of cell division followed by a phase of cell elongation35,60,88. However, recent evidence suggests a diversity of molecular mechanisms recruited to achieve this, with KNOX and TCP4 genes respectively proposed as regulators of spur development in toadflaxes (Linaria, Plantaginaceae)60 and columbines (Aquilegia, Ranunculaceae)61.

Macro–microevolution transition. Recent speciation potentially driven by nectar spurs has been studied mainly in the North American clade of the genus Aquilegia. In this lineage, changes in spur length are associated with shifts in the main pollinators (bees, hummingbirds or hawkmoths)57, and spur length differences contribute to the reproductive isolation between co-occurring species89. Ontogenetically, variation in spur length across species is the result of changes in the duration of the phase of cell elongation during spur development35. It remains to be determined whether this heterochronic mechanism of spur length evolution also characterizes other unrelated spurred lineages.

Microevolutionary scale. Provides the best opportunities to test evolutionary models of spur length change, such as the coevolutionary race and pollinator shift hypotheses57, and the degree to which spur evolution contributes to incipient divergence. In Scandinavian populations of the orchid Platanthera bifolia, geographic variation in spur length is correlated with the proboscis length of distinct local pollinators, suggesting that intraspecific changes in spur length are driven by pollinator shifts90. The developmental mechanisms behind these changes, as well as their potential contribution to incipient divergence and eventual speciation, remain to be studied.

Model systems to study the evolution of nectar spurs. Ordinal-level phylogeny of flowering plants, with red branches indicating orders in which nectar spurs have evolved87. Three phylogenetically disparate genera in which evolution of nectar spurs is being investigated at different scales are shown. a, Aquilegia: species with different pollination syndromes display contrasting spur lengths, as exemplified by A. sibirica (approx. 10 mm, bee-pollinated, top), A. formosa (approx. 20 mm, hummingbird-pollinated, centre) and A. chrysantha (approx. 70 mm, hawkmoth-pollinated, bottom). b, Linaria: L. salzmannii (approx. 13 mm, left) and L. clementei (approx. 3 mm, right) are two closely related species with nearly identical floral morphology but contrasting spur lengths. c, Platanthera: intraspecific variation in spur length in P. bifolia (approx. 15–40 mm) correlates with the proboscis length of local pollinators. Nectar spurs are indicated by arrow heads in all photos. Photos reproduced with permission from E. S. Ballerini (a), M. Fernández-Mazuecos (b) and J. Quiles (c).

In addition to key innovations, highly labile traits may be involved in numerous speciation events throughout plant lineages. The best example is probably flower colour, a plant trait with some of the best-understood developmental pathways, particularly those related to anthocyanin pigmentation22. Flower colour is frequently involved in pollinator-driven speciation events in distant angiosperm families23, providing another useful example of phylogenetic replication at the macroevolutionary scale. In this case, continuing work in multiple models (Antirrhinum, Petunia and Aquilegia, among others; Fig. 1a,b; Box 1) reveals that evolutionary changes tend to occur in ‘optimally pleiotropic’ components of developmental pathways, that is, those that maximize change in the trait under selection while at the same time minimizing deleterious effects on other traits, such as seed coat development, UV resistance and pathogen defence19,24.

Figure 1: Model systems for the evolutionary developmental study of plant speciation.
Figure 1

a, Antirrhinum spp.: variation in flower and leaf morphology in a recent radiation, represented by A. majus (left), A. braun-blanquetii (centre) and A. charidemi (right). b, Petunia spp.: transitions in flower tube length, stigma exsertion and colour associated with pollinator shifts, represented by P. inflata (bee-pollinated, left), P. exserta (hummingbird-pollinated, centre) and P. axillaris (moth-pollinated, right). c, Mimulus spp.: differences in floral morphology between two sister species with contrasting pollination syndromes, M. lewisii (bee-pollinated, left) and M. cardinalis (hummingbird-pollinated, right). d, Tragopogon spp.: capitula of the allopolyploid hybrid T. mirus (centre) and its parent species T. dubius (left) and T. porrifolius (right). Photos reproduced with permission from A. Hudson (a), H. Sheehan (b), H. D. Bradshaw (c), E. Mavrodiev (d, left and centre) and A. N. Doust (d, right).

The role of the fossil record. A source of macroevolutionary information that is frequently overlooked by plant evo-devo is the fossil record. Fossils provide invaluable information about evolutionary patterns that cannot be inferred from extant lineages alone, including calibrations for phylogenetic dating analyses to help estimate the timing of developmental evolution25. A detailed study of the fossil record also allows us to assess historical changes in the disparity (morphological variety) of clades, and their correlation, or lack thereof, with changes in diversity (number of species). Although only preliminary analyses are available for plants26, diversity and disparity seem to be fundamentally decoupled, with maximum disparity frequently being achieved early during diversification of a clade. This is usually followed by an increase in diversity that is not accompanied by further increases in disparity. Proposed explanations for this pattern of diversification through small variations on early evolving themes, also found in animals, include developmental constraints and ecological restrictions. An eco-evo-devo approach integrating the study of fossil and extant lineages may help to distinguish these two non-mutually exclusive hypotheses.

Remarkably, fossils can preserve diagnostic features of plant development, and even structural evidence for developmental regulatory mechanisms of extinct plants27,​28,​29, thus potentially containing information on the origin of evolutionary innovations important for diversification. This is particularly relevant for ancient evolutionary transitions whose signature in extant lineages may be limited. Examples are provided by the evolution of woody growth and tree architecture across plant lineages. Among extant plant lineages, wood is only produced by seed plants. However, the fossil record shows that woody growth has evolved several times in the course of vascular plant diversification, and there is strong developmental evidence that these multiple acquisitions were mediated by a common polar auxin regulatory pathway29,30. Evolution of tree architecture is particularly intriguing in the arborescent lycopsids (Lepidodendrales), which speciated profusely in Carboniferous coal-swamp forests and whose closest living relatives are the herbaceous quillworts (Isoetes). Interestingly, both Lepidodendrales and Isoetes share a pattern of bipolar growth in which the rooting system is a highly modified shoot system known as the rhizomorph29. This indicates that developmental studies of Isoetes and comparison with other living lycopsids can provide insights into the development and diversification of the long-extinct arborescent lycopsids31.

The examples above show that macroevolutionary studies provide invaluable information on large-scale patterns of speciation and developmental evolution. However, they do not provide details of the mechanisms involved in particular speciation events. For that, study systems at finer evolutionary scales (closely related species and populations) are required.

The macro–microevolution transition

At the macro–microevolution transition, where separation of closely related species is studied, developmental repatterning can underlie the origin of reproductive barriers. It also plays a key role in generating further phenotypic changes that distinguish closely related species (Fig. 1). All possible types of developmental repatterning may be involved in these speciation events, including changes in timing, spatial distribution, quantity and type of developmental activities at each of the molecular, cellular and organismal levels32. Examples of developmental repatterning between closely related species (directly involved in reproductive isolation or not) include shifts in flowering time33, inflorescence architecture34, nectar spur length35, flower colour36, petal cell shape37 and leaf shape38. Repatterning of multiple developmental traits frequently occurs associated with a speciation event. The most conspicuous examples involve shifts in pollination syndromes, commonly studied through the developmental and genetic comparison of closely related species with contrasting pollinators (for example, in Mimulus39, Petunia40, Aquilegia41; Fig. 1b,c; Box 1).

Shifts in pollination syndrome. Perhaps the best-studied system in which evo-devo research has shed light on the separation of closely related species is the solanaceous genus Petunia. The 20 species of Petunia originated in South America, and their radiation is considered to have occurred within the past three million years42. Major phenotypic differences between species are mostly related to the flower, and these differences underpin divergent relationships with different pollinating animals43. Specifically, attention has focused on understanding the differences in corolla tube length, stigma exsertion, anthocyanin production and UV-absorbing flavonol content that distinguish bee-pollinated species such as P. integrifolia and P. inflata (short tube, no stigma exsertion, anthocyanin, no flavonols), moth-pollinated species such as P. axillaris (long tube, no stigma exsertion, no anthocyanin, flavonols) and the single hummingbird-pollinated species P. exserta (long tube, stigma exsertion, anthocyanin, no flavonols) (Fig. 1b). These studies have been facilitated by the ability to cross these different species and by the development of a range of genetic, genomic and transgenic resources. By combining these approaches to isolate individual traits of the different pollination syndromes, and by using pollinator behaviour studies, it has been possible, for several of these characters, to identify both the molecular basis of trait repatterning and the consequences for pollinator behaviour and reproductive isolation40,44,45. These combined studies in a single system have revealed novel conceptual insights into the developmental shifts underpinning ecological speciation. One such insight is the discovery that many of the molecular changes target transcriptional regulators of developmental pathways, with the R2R3-MYB family of transcription factors being a key target in Petunia. These proteins can be thought of as mid-level control points, downstream of the essential regulation of floral organ identity but specifying the shape, pattern and colour of those organs by direct activation of structural genes encoding enzymes and cytoskeletal components46. A second insight is that multiple molecular evolutionary events may underpin the same phenotypic change if that change is of sufficient selective advantage. Hoballah et al.40 reported that loss of function of AN2, a MYB regulator of anthocyanin production, had occurred at least five times independently in wild-sampled P. axillaris, a white-flowered moth-pollinated species.

Perhaps most striking, though, has been the unexpected discovery that many of the genes controlling traits involved in pollinator specificity have become linked in Petunia, generating a multigene ‘speciation locus’ (or ‘speciation island’) on chromosome II47. This is a novel feature of Petunia — the same genes regulating anthocyanin production, UV absorption, male and female reproductive organ position and scent are distributed across multiple chromosomes in other solanaceous species. It is likely that this clustering of key genes promotes linkage disequilibrium and avoids pollination syndromes being disturbed by recombination that could reduce fitness.

Reproductive isolation in other ways. The Petunia system emphasizes the importance of multiple trait repatterning to ensure reproductive isolation through differential pollination syndromes. However, single aspects of flower development can also diverge between closely related species, generating reproductive isolation more simply. One classic example is the divergence of flowering time between closely related species growing in the same habitat. This displacement of flowering phenology has been observed in multiple systems under different conditions, and is particularly striking when repeated in multiple different habitats (for example, Lobo et al.48 studying flowering time of bat-pollinated Bombacaceae in three different habitats with different rainfall patterns). Displacement of flowering can occur to minimize competition for pollinator attention and interspecific hybridization between established species, or it may occur as part of the process of reproductive isolation as species diverge. Ellis et al.33 observed displacement of flowering time over a 14-week winter rainfall season for species of the stone plant Argyroderma, and interpreted this displacement as an adaptation to isolate populations that had diverged in their tolerance for different soil conditions, facilitating full speciation. The processes that determine when a plant flowers are well described in Arabidopsis thaliana, with multiple environmental and endogenous pathways converging on the activity of a set of floral meristem identity genes (reviewed by Holt et al.49 and Glover50). To fully understand the molecular basis of the developmental transitions in flowering time, observed in various plant radiations, multi-species studies will need to be connected to the intraspecific work currently exploring variation in flowering time in different ecotypes of Arabidopsis51,​52,​53.

Reproductive isolation between close relatives can also result from shifts in breeding system. Shifts from outcrossing to selfing in flowering plants are commonly associated with the evolution of a set of phenotypic traits known as the ‘selfing syndrome’: smaller flowers, reduced pollen production and loss of scent and nectar production54. The developmental changes producing the selfing syndrome are being studied in the sister species Capsella grandiflora (outcrossing, large flowers) and C. rubella (selfing, small flowers). Reduced petal size in C. rubella results from a reduction in the number of petal cells caused by a shortening of the cell division period. Allelic variation in the intron of a general growth regulator, affecting the levels of STERILE APETALA (SAP) protein in developing petals, has contributed to this change55. Interestingly, it seems that the small-petal allele of SAP was already present in the ancestral outcrossing population, explaining the rapid evolutionary reduction of petal size during speciation. In addition, C. rubella has lost a major component of floral scent present in C. grandiflora (benzaldehyde) as a result of repeated inactivations of the CNL1 gene (encoding the enzyme cinnamate:CoA ligase), caused by independent mutations in its coding sequence56.

Another example of reproductive isolation generated through an evolutionary change to a developmental programme is the specialization of plant species on pollinators with particular lengths of feeding apparatus, through transitions in the length of floral tubes and nectar spurs (Box 1). This sort of change can be associated with major shifts of pollinator, and therefore with other changes to flower morphology and colour, or it can occur in isolation of other traits and simply select between insects of different proboscis length. In a classic study of nectar spur evolution in North American species of Aquilegia, Whittall and Hodges57 demonstrated that the evolution of increasingly long spurs is driven by speciation events involving shifts between pollinators with increasingly long mouth parts (bees, hummingbirds and hawkmoths). Other studies have focused on simpler transitions in spur length between closely related species of orchid58 and Linaria59. Current advances in our understanding of nectar spur development are allowing analysis of the molecular evolutionary processes underpinning these speciation events35,60,61 (see Box 1).

Speciation by hybridization and polyploidization. Hybridization and polyploidy can rapidly generate reproductive isolation and therefore lead to speciation in plants62. In the genus Tragopogon, for example, new allopolyploid species (T. miscellus, T. mirus) have evolved in the last century in North America as a result of hybridization between three naturalized Eurasian species (Fig. 1d). Each allopolyploid has been produced multiple times in independent hybridization events, and they can also be generated synthetically, providing excellent opportunities for comparative analysis. Upon allopolyploidization, genes inherited from the progenitor species can be differently expressed, silenced or even lost63. This may lead to developmental variation, such as that found between populations of T. miscellus, which display long or short ligules depending on the identity of the maternal and paternal parents.

Taxonomically diagnostic traits. While it is tempting to focus on traits directly involved in reproductive isolation, phenotypic changes resulting from developmental repatterning are frequently used to taxonomically delimit species, even if their involvement in the initial stages of reproductive isolation is uncertain. One example is evolution of leaf shape, which has been studied in the genera Antirrhinum38 (Fig. 1a) and Solanum64. Antirrhinum comprises around 25 species, originating approximately four million years ago in the Mediterranean region65. Analysis of QTLs associated with leaf size and shape following crosses between small-leaved and large-leaved species suggested that the species had diverged in response to fluctuating selection regimes, consistent with a radiation in a period of climate and vegetation cycles38. Leaf size and shape are understood to be products of the combined amount of cell division and cell expansion that occurs throughout organ development, and these processes are controlled in a coordinate way. Molecular evolution of these processes is often developmentally constrained, causing leaves and other organs to evolve together38, generating major phenotypic differences that can be used in taxonomic species description as well as underpinning selection in different environmental conditions. In Solanum, closely related species display contrasting levels of complexity of compound leaves. Variation in expression of the BLADE-ON-PETIOLE (BOP) transcription factor seems to explain this diversity through dynamic rewiring of interactions in the gene regulatory network for leaf development64. This includes the alteration of the transcript levels of KNOX genes, which have been recurrently recruited to generate leaf diversity during plant evolution.

Microevolutionary scale

The examples in the previous section illustrate evolutionary developmental changes involving recently diversified species, usually with reproductive barriers already in place. To better understand how traits involved in speciation first evolve, developmental variation can be investigated at an even finer scale, within species and populations.

Genetically based intraspecific variation. Relatively little attention has been paid to microevolution in the plant evo-devo literature. Microevolutionary processes have been traditionally studied from two interacting perspectives: (1) population genetics, including the study of genetic variation in populations and allele frequency changes due to mutation, selection, migration and drift; and (2) evolutionary ecology, which investigates the biotic and abiotic interactions underlying the selective pressures that lead to evolutionary change in populations. A deeper understanding of plant speciation emerges from the integration of these approaches with developmental biology. The evo-devo approach to microevolution (micro-evo-devo5) examines evolvability, the ability of species and populations to produce heritable phenotypic variation, as determined by genetic architecture and developmental constraints6,66. In this way, it provides insight into the processes that supply the raw material for adaptation, evolution and speciation5,67. This generally involves the study of developmental variation across populations of the same species and within populations, particularly those polymorphisms that may underlie local adaptation, divergence between populations and, potentially, the establishment of reproductive barriers (Fig. 2).

Figure 2: Species showing intraspecific variation in developmental traits relevant to plant speciation.
Figure 2

a, Mimulus aurantiacus: floral colour variation associated with pollinator preferences between a red-flowered ecotype (preferred by hummingbirds, left) and a yellow-flowered ecotype (preferred by hawkmoths, right). b, Erysimum mediohispanicum: variation in floral symmetry correlated with fitness differences, including radial (left), dissymmetric (centre) and zygomorphic (right) flowers. c, Gorteria diffusa: variation in presence of petal spots and degree of sexual deception between three morphotypes (from left to right: Steinkopf, Cal and Buffels). Photos reproduced with permission from M. A. Streisfield (a), J. M. Gómez (b) and G. Mellers (c).

A fertile field for microevolutionary research in plants is the study of flower colour polymorphisms. Flower pigmentation is involved in pollinator specialization23,68, and its molecular and developmental basis has been well studied22,24. In addition, intraspecific colour polymorphisms are relatively common in nature, and they frequently involve few genetic changes. A link between intraspecific flower colour variation and incipient diversification has been demonstrated in the sticky monkeyflower (Mimulus aurantiacus). A number of studies69,​70,​71 have addressed the genetic basis of flower colour variation, population genetics, pollinator interactions and isolating barriers in two closely related morphs of M. aurantiacus distributed in southwestern California: a red-flowered ecotype preferentially pollinated by hummingbirds and a yellow-flowered one preferred by hawkmoths (Fig. 2a). This multi-disciplinary approach has revealed incipient ecological speciation in the face of gene flow, primarily resulting from pollinator preferences causing divergent selection on an R2R3-MYB transcription factor (MaMyb2) involved in the regulation of the anthocyanin biosynthetic pathway71. A cis-regulatory change in MaMyb2 is responsible for the colour change underlying incipient pre-mating isolation between ecotypes. This example reveals microevolutionary mechanisms by which traits involved in pollination syndrome shifts first evolve.

Another trait relevant to speciation that is amenable to population-level research is floral symmetry. Floral zygomorphy is known to have evolved in several genera of Brassicaceae in correlation with differences in expression of CYC2 during corolla development72. To understand how zygomorphy has evolved at a microevolutionary scale, an ideal system is provided by Erysimum mediohispanicum, a member of the Brassicaceae displaying heritable intraspecific variation in floral symmetry, from actinomorphic to zygomorphic73 (Fig. 2b). Evolutionary ecological approaches show that plants bearing zygomorphic flowers have the highest fitness, and that strong selection on corolla shape is exerted by pollinators74. By analyzing the developmental genetic basis of floral symmetry variation in E. mediohispanicum, the microevolutionary process by which zygomorphy evolves would be more fully understood, and this would in turn enhance our understanding of macroevolutionary patterns in Brassicaceae.

Many other traits potentially involved in speciation are being investigated using polymorphic target species, including the following examples: flowering time differentiation between locally adapted populations in Arabidopsis thaliana51; continuous variation in pollination by sexual deception in the South African beetle daisy (Gorteria diffusa; Fig. 2c)75; the recurrent parallel divergence of morphologically distinct ecotypes adapted to contrasting habitats in the Australian groundsel Senecio lautus76; and adaptive variation in the production of leaf trichomes, involved in resistance to herbivory, in Arabidopsis lyrata77. There is the exciting potential for the comparative microevolutionary study of similar traits in distant lineages to provide metamodels linking the macro- and microevolutionary scales.

Beyond genetic variation. While intraspecific phenotypic variation discussed thus far is considered to be the result of genetic changes, recent research has highlighted a potential role in speciation for other components of variation. Phenotypic plasticity, the capacity of a genotype to produce alternative phenotypes in response to environmental variation, has been suggested as a facilitator of adaptive divergence and speciation78,79. Intraspecific phenotypic differences initially generated by plasticity may be fixed in different populations by natural selection in the process of genetic assimilation, and can then contribute to potentially rapid genetic divergence, reproductive isolation and eventually speciation78. As a result, the developmental mechanisms responsible for plasticity may parallel those underlying interspecific diversity. For example, heterophylly in the North American lake cress (Rorippa aquatica), involving morphological differences between leaves developing under submerged and terrestrial conditions, is the result of environmentally induced changes in the expression of KNOX1 genes, which are also implicated in the diversification of leaf shape across species of the same family80.

Related to phenotypic plasticity, there is speculation that heritable epigenetic variation, shaped by the environment and natural selection, might also aid evolutionary change and speciation81,​82,​83. Epigenetic diversity, triggered by environmental changes, may enable genetically depauperate populations to quickly adapt until genetic assimilation fixes phenotypic differences. Interestingly, one of the first naturally occurring morphological mutants to be genetically characterized, the peloric mutant of Linaria vulgaris (showing radially symmetrical flowers instead of the zygomorphic flowers that are characteristic of Linaria), was found to be an epimutant resulting from extensive methylation of the CYC gene84. Although the evolutionary significance of this particular mutant is probably limited given its compromised reproductive success, evidence has since been found of heritable intraspecific epigenetic variation correlated with phenotypic differences and potentially subject to natural selection85. For example, in the yellow monkeyflower (Mimulus guttatus), epigenetically inherited variation in trichome density is induced by herbivore damage, and is correlated with differential regulation of a MYB MIXTA-like transcription factor86. The emerging field of population epigenetics, combined with ecological and developmental approaches, can provide insights into this still largely hypothetical link between epigenetics and speciation.

Concluding remarks

The processes of speciation have puzzled evolutionary biologists for over 150 years. Many models to explain how new species emerge have been proposed and many systems developed in which to test those models. It is clear that speciation events result from a combination of multiple molecular, environmental and stochastic factors. However, the recent input of evolutionary developmental biology into this field has generated new insights. It has allowed both the crystallization of novel concepts surrounding speciation processes and the revisiting of old questions (Box 2). We conclude that a strong interaction between developmental biology and evolutionary biology (including phylogenetics, population genetics, evolutionary ecology, palaeontology; Box 3) is crucial to retain momentum in the drive to gain a deeper understanding of plant speciation.

Box 2: Hopeful monsters.

The evolutionary relevance of large-effect mutations in evolution and speciation is at the centre of a long-standing debate in evolutionary biology10. Although widely accepted evolutionary models regard gradual change as the most likely mode of evolution, it has been frequently argued by developmental biologists that homeotic mutations, changing the identity of whole organs, may have played a role in some major evolutionary transitions2,3. For example, evolutionary changes in floral organ identity have been hypothesized to be the result of homeotic mutations involving changes in the expression domains of genes in the ABC model of flower development91. While this hypothesis is intriguing, systems in which the feasibility of such changes can be studied at the microevolutionary scale are required to test it. The Stamenoid petals (Spe) mutant of the shepherd's purse (Capsella bursa-pastoris) has been proposed as a suitable model for this92. It is a naturally occurring floral homeotic mutant in which petals have been transformed into stamens. Unlike other known homeotic mutants, it forms stable populations in the wild, mixed with wild-type plants. The mutation has been shown to be heritable and involves a single locus, hypothesized to be a class C floral identity gene93. Both morphs seem to have similar fitness, and a degree of genetic differentiation and reproductive isolation between them has been detected in a German locality, suggesting incipient speciation94,95. Even if considered a rarity, this system nicely bridges microevolutionary processes and macroevolutionary outcomes, and hints at the feasibility of saltational changes giving rise to ‘hopeful monsters’ of potential long-term evolutionary relevance96.

A hypothetical hopeful monster. Flowers of wild-type Capsella bursa-pastoris (left) and the naturally occurring Spe mutant of the same species (right). Photos reproduced with permission from G. Theißen.

Box 3: The need for a robust phylogenetic context.

The role of phylogenetics in evolutionary developmental biology has been highlighted since the origins of evo-devo97, and it is particularly crucial when the focus is on speciation. Indeed, phylogenetic relationships have to be known if the sequence and direction of developmental changes in the course of speciation are to be understood98, including, for example, the detection of instances of parallelism that may result from developmental biases. However, integration of phylogenetic and developmental data is often lacking, and the use of new analytic tools to achieve it is desirable99,100. In addition, speciation studies frequently involve recently diverged species or populations whose phylogenetic relationships cannot be easily resolved using conventional phylogenetic approaches. To that end, high-throughput sequencing methods capable of providing genome-wide markers are required. In their study of flower colour divergence during incipient diversification in the Mimulus aurantiacus complex, for example, Stankowski and Streisfeld70 provide a good example of the use of a robust phylogenetic framework, developed using RAD-Seq markers, to reconstruct evolutionary developmental changes. According to phylogenetic analyses, red flowers have been acquired in two independent lineages with yellow flowered ancestors. In both cases, the red pigmentation is the result of a cis-regulatory mutation in the gene MaMyb2. Interestingly, population genetic analyses suggest that a single red allele may have evolved and subsequently been transferred between the two red-flowered morphs by introgression.

Additional information

How to cite this article: Fernández-Mazuecos, M. & Glover, B. J. The evo-devo of plant speciation. Nat. Ecol. Evol. 1, 0110 (2017).

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.


  1. 1.

    & Speciation (Sinauer Associates, 2004).

  2. 2.

    Evolution 2nd edn (Sinauer Associates, 2009).

  3. 3.

    Evolution: a Developmental Approach (Wiley-Blackwell, 2011).

  4. 4.

    , & MADS-domain transcription factors and the floral quartet model of flower development: linking plant development and evolution. Development 143, 3259–3271 (2016).

  5. 5.

    , , & A perspective on micro-evo-devo: progress and potential. Genetics 195, 625–634 (2013).

  6. 6.

    & On the evolutionary developmental biology of speciation. Evol. Biol. 39, 242–254 (2012).

  7. 7.

    & Plant speciation. Science 317, 910–914 (2007).

  8. 8.

    & Speciation genes in plants. Ann. Bot. 106, 439–455 (2010).

  9. 9.

    & Associations between floral specialization and species diversity: cause, effect, or correlation?. Evol. Ecol. 23, 159–179 (2009).

  10. 10.

    in Contemporary Debates in Philosophy of Biology (eds & ) 169–179 (Wiley-Blackwell, 2010).

  11. 11.

    in Contemporary Debates in the Philosophy of Biology (eds & ) 180–193 (Wiley-Blackwell, 2010).

  12. 12.

    & The significance of developmental robustness for species diversity. Ann. Bot. 117, 725–732 (2016).

  13. 13.

    & Key evolutionary innovations and their ecological mechanisms. Hist. Biol. 10, 151–173 (1995).

  14. 14.

    et al. in Ecology and Evolution of Flowers (eds Harder, L. D. & Barrett, S. C. H.) 311–325 (Oxford Univ. Press, 2006).

  15. 15.

    et al. Polyploidy and angiosperm diversification. Am. J. Bot. 96, 336–348 (2009).

  16. 16.

    et al. Ancestral polyploidy in seed plants and angiosperms. Nature 473, 97–100 (2011).

  17. 17.

    , & Ancient whole genome duplications, novelty and diversification: the WGD Radiation Lag-Time Model. Curr. Opin. Plant Biol. 15, 147–153 (2012).

  18. 18.

    The co-radiations of pollinating insects and angiosperms in the Cretaceous. Ann. Missouri Bot. Gard. 86, 373–406 (1999).

  19. 19.

    Metamodels and phylogenetic replication: a systematic approach to the evolution of developmental pathways. Evolution 63, 2771–2789 (2009).

  20. 20.

    Floral symmetry affects speciation rates in angiosperms. Proc. R. Soc. Lond. B 271, 603–608 (2004).

  21. 21.

    Bilateral flower symmetry—how, when and why? Curr. Opin. Plant Biol. 17, 146–152 (2014).

  22. 22.

    The genetics and biochemistry of floral pigments. Annu. Rev. Plant Biol. 57, 761–780 (2006).

  23. 23.

    Evolutionary transitions in floral color. Int. J. Plant Sci. 169, 7–21 (2008).

  24. 24.

    & Lessons from flower colour evolution on targets of selection. J. Exp. Bot. 63, 5741–5749 (2012).

  25. 25.

    Written in stone: fossils, genes and evo–devo. Nat. Rev. Genet. 8, 911–920 (2007).

  26. 26.

    , , & Why should we investigate the morphological disparity of plant clades?. Ann. Bot. 117, 859–879 (2015).

  27. 27.

    The evolution of plant development in a paleontological context. Curr. Opin. Plant Biol. 13, 102–107 (2010).

  28. 28.

    , & Unique cellular organization in the oldest root meristem. Curr. Biol. 26, 1629–1633 (2016).

  29. 29.

    , & Plant evolution at the interface of paleontology and developmental biology: An organism-centered paradigm. Am. J. Bot. 101, 899–913 (2014).

  30. 30.

    , , & A fossil record for growth regulation: the role of auxin in wood evolution. Ann. Missouri Bot. Gard. 95, 121–134 (2008).

  31. 31.

    & Basal meristem and root development in Isoetes asiatica and Isoetes japonica. Int. J. Plant Sci. 162, 1225–1235 (2001).

  32. 32.

    The emerging conceptual framework of evolutionary developmental biology. Nature 415, 757–764 (2002).

  33. 33.

    , & Evolutionary radiation of “stone plants” in the genus Argyroderma (Aizoaceae): unraveling the effects of landscape, habitat, and flowering time. Evolution 60, 39–55 (2006).

  34. 34.

    A cladistic analysis of species groups in Weinmannia (Cunoniaceae) based on morphology and inflorescence architecture. Ann. Missouri Bot. Gard. 85, 565–593 (1998).

  35. 35.

    , , , & Evolution of spur-length diversity in Aquilegia petals is achieved solely through cell-shape anisotropy. Proc. R. Soc. Lond. B 279, 1640–1645 (2012).

  36. 36.

    et al. Evolutionary paths underlying flower color variation in Antirrhinum. Science 313, 963–966 (2006).

  37. 37.

    et al. Comparative micromorphology of petals in Macaronesian Lotus (Leguminosae) reveals a loss of papillose conical cells during the evolution of bird pollination. Int. J. Plant Sci. 173, 365–374 (2012).

  38. 38.

    et al. Evolution of allometry in Antirrhinum. Plant Cell 21, 2999–3007 (2009).

  39. 39.

    & Allele substitution at a flower colour locus produces a pollinator shift in monkeyflowers. Nature 426, 176–178 (2003).

  40. 40.

    et al. Single gene-mediated shift in pollinator attraction in Petunia. Plant Cell 19, 779–790 (2007).

  41. 41.

    , , & Genetics of floral traits influencing reproductive isolation between Aquilegia formosa and Aquilegia pubescens. Am. Nat. 159, S51–S60 (2002).

  42. 42.

    et al. Multilocus phylogeny reconstruction: new insights into the evolutionary history of the genus Petunia. Mol. Phylogenet. Evol. 81, 19–28 (2014).

  43. 43.

    , , & in Petunia: Evolutionary, Developmental and Physiological Genetics (eds Gerats, T. & Strommer, J.) 29–49 (Springer-Verlag, 2009).

  44. 44.

    et al. Pollinator choice in Petunia depends on two major genetic loci for floral scent production. Curr. Biol. 21, 730–739 (2011).

  45. 45.

    et al. MYB-FL controls gain and loss of floral UV absorbance, a key trait affecting pollinator preference and reproductive isolation. Nat. Genet. 48, 159–166 (2016).

  46. 46.

    , & The genetic control of flower–pollinator specificity. Curr. Opin. Plant Biol. 16, 422–428 (2013).

  47. 47.

    et al. Tight genetic linkage of prezygotic barrier loci creates a multifunctional speciation island in Petunia. Curr. Biol. 23, 873–877 (2013).

  48. 48.

    et al. Factors affecting phenological patterns of bombacaceous trees in seasonal forests in Costa Rica and Mexico. Am. J. Bot. 90, 1054–1063 (2003).

  49. 49.

    , , & Signaling in shoot and flower meristems of Arabidopsis thaliana. Curr. Opin. Plant Biol. 17, 96–102 (2014).

  50. 50.

    Understanding flowers and flowering: an integrated approach (Oxford Univ. Press, 2014).

  51. 51.

    , , , & Altitudinal and climatic adaptation is mediated by flowering traits and FRI, FLC, and PHYC genes in Arabidopsis. Plant Physiol. 157, 1942–1955 (2011).

  52. 52.

    , , , & Genetic architecture of flowering time differentiation between locally adapted populations of Arabidopsis thaliana. New Phytol. 197, 1321–1331 (2013).

  53. 53.

    et al. Variation in Arabidopsis flowering time associated with cis-regulatory variation in CONSTANS. Nat. Commun. 5, 3651 (2014).

  54. 54.

    & The selfing syndrome: a model for studying the genetic and evolutionary basis of morphological adaptation in plants. Ann. Bot. 107, 1433–1443 (2011).

  55. 55.

    et al. Standing genetic variation in a tissue-specific enhancer underlies selfing-syndrome evolution in Capsella. Proc. Natl Acad. Sci. USA 113, 13911–13916 (2016).

  56. 56.

    et al. Repeated inactivation of the first committed enzyme underlies the loss of benzaldehyde emission after the selfing transition in Capsella. Curr. Biol. 26, 3313–3319 (2016).

  57. 57.

    & Pollinator shifts drive increasingly long nectar spurs in columbine flowers. Nature 447, 706–709 (2007).

  58. 58.

    , , & Floral ontogenetic evidence of repeated speciation via paedomorphosis in subtribe Orchidinae (Orchidaceae). Bot. J. Linn. Soc. 157, 429–454 (2008).

  59. 59.

    et al. Bees explain floral variation in a recent radiation of Linaria. J. Evolution. Biol. 28, 851–863 (2015).

  60. 60.

    , , , & Characterization of Linaria KNOX genes suggests a role in petal-spur development. Plant. J. 68, 703–714 (2011).

  61. 61.

    , , , & Molecular basis for three-dimensional elaboration of the Aquilegia petal spur. Proc. R. Soc. Lond. B 282, 20142778 (2015).

  62. 62.

    & The role of hybridization in plant speciation. Annu. Rev. Plant Biol. 60, 561–588 (2009).

  63. 63.

    et al. The legacy of diploid progenitors in allopolyploid gene expression patterns. Phil. Trans. R. Soc. B 369, 20130354 (2014).

  64. 64.

    et al. Evolutionary developmental transcriptomics reveals a gene network module regulating interspecific diversity in plant leaf shape. Proc. Natl Acad. Sci. USA 111, E2616–E2621 (2014).

  65. 65.

    , , , & A geographical pattern of Antirrhinum (Scrophulariaceae) speciation since the Pliocene based on plastid and nuclear DNA polymorphisms. J. Biogeogr. 36, 1297–1312 (2009).

  66. 66.

    Is evolvability evolvable?. Nat. Rev. Genet. 9, 75–82 (2008).

  67. 67.

    The micro-evolution of development. Genetica 129, 1–5 (2007).

  68. 68.

    , , , & Pollination syndromes and floral specialization. Annu. Rev. Ecol. Evol. Systemat. 35, 375–403 (2004).

  69. 69.

    & Strong premating reproductive isolation drives incipient speciation in Mimulus aurantiacus. Evolution 69, 447–461 (2015).

  70. 70.

    & Introgressive hybridization facilitates adaptive divergence in a recent radiation of monkeyflowers. Proc. R. Soc. Lond. B 282, 20151666 (2015).

  71. 71.

    , & Divergent selection drives genetic differentiation in an R2R3-MYB transcription factor that contributes to incipient speciation in Mimulus aurantiacus. PLoS Genet. 9, e1003385 (2013).

  72. 72.

    , , , & Corolla monosymmetry: evolution of a morphological novelty in the Brassicaceae family. Mol. Biol. Evol. 29, 1241–1254 (2012).

  73. 73.

    , , & Heritability and genetic correlation of corolla shape and size in Erysimum mediohispanicum. Evolution 63, 1820–1831 (2009).

  74. 74.

    , & Natural selection on Erysimum mediohispanicum flower shape: insights into the evolution of zygomorphy. Am. Nat. 168, 531–545 (2006).

  75. 75.

    et al. Floral trait variation and integration as a function of sexual deception in Gorteria diffusa. Phil. Trans. R. Soc. Lond. B 369, 20130563 (2014).

  76. 76.

    et al. Convergence and divergence during the adaptation to similar environments by an Australian groundsel. Evolution 67, 2515–2529 (2013).

  77. 77.

    , , , & Gene, phenotype and function: GLABROUS1 and resistance to herbivory in natural populations of Arabidopsis lyrata. Mol. Ecol. 16, 453–462 (2007).

  78. 78.

    et al. Phenotypic plasticity's impacts on diversification and speciation. Trends Ecol. Evol. 25, 459–467 (2010).

  79. 79.

    & Evaluating ‘plasticity-first’ evolution in nature: key criteria and empirical approaches. Trends Ecol. Evol. 31, 563–574 (2016).

  80. 80.

    et al. Regulation of the KNOX-GA gene module induces heterophyllic alteration in North American lake cress. Plant Cell 26, 4733–4748 (2014).

  81. 81.

    , , & Environmental heterogeneity and phenotypic divergence: can heritable epigenetic variation aid speciation?. Genet. Res. Int. 2012, 698421 (2012).

  82. 82.

    Epigenetic responses to environmental change and their evolutionary implications. Phil. Trans. R. Soc. Lond. B 364, 3403–3418 (2009).

  83. 83.

    et al. Stable epigenetic effects impact adaptation in allopolyploid orchids (Dactylorhiza: Orchidaceae). Mol. Biol. Evol. 27, 2465–2473 (2010).

  84. 84.

    , & An epigenetic mutation responsible for natural variation in floral symmetry. Nature 401, 157–161 (1999).

  85. 85.

    & Epigenetic differentiation and relationship to adaptive genetic divergence in discrete populations of the violet Viola cazorlensis. New Phytol. 187, 867–876 (2010).

  86. 86.

    , , , & Differential regulation of a MYB transcription factor is correlated with transgenerational epigenetic inheritance of trichome density in Mimulus guttatus. New Phytol. 191, 251–263 (2011).

  87. 87.

    Floral nectar spurs and diversification. Int. J. Plant Sci. 158, 81–88 (1997).

  88. 88.

    & The relationship between cell division and elongation during development of the nectar-yielding petal spur in Centranthus ruber (Valerianaceae). Ann. Bot. 115, 641–649 (2015).

  89. 89.

    & Floral isolation between Aquilegia formosa and Aquilegia pubescens. Proc. R. Soc. Lond. B 266, 2247–2252 (1999).

  90. 90.

    et al. Pollinator shifts and the evolution of spur length in the moth-pollinated orchid Platanthera bifolia. Ann. Bot. 113, 267–275 (2014).

  91. 91.

    Saltational evolution: hopeful monsters are here to stay. Theory Biosci. 128, 43–51 (2009).

  92. 92.

    et al. Catching a ‘hopeful monster’: shepherd's purse (Capsella bursa-pastoris) as a model system to study the evolution of flower development. J. Exp. Bot. 57, 3531–3542 (2006).

  93. 93.

    , , & Mapping a floral trait in Shepherds purse–'stamenoid petals’ in natural populations of Capsella bursa-pastoris (L.) Medik. Flora 208, 641–647 (2013).

  94. 94.

    , & Genetic differentiation and reproductive isolation of a naturally occurring floral homeotic mutant within a wild-type population of Capsella bursa-pastoris (Brassicaceae). Mol. Ecol. 18, 2659–2667 (2009).

  95. 95.

    et al. Floral visitation and reproductive traits of Stamenoid petals, a naturally occurring floral homeotic variant of Capsella bursa-pastoris (Brassicaceae). Planta 230, 1239–1249 (2009).

  96. 96.

    Revenge of the hopeful monster. Nature 463, 864–867 (2010).

  97. 97.

    Ontogeny and Phylogeny (Harvard Univ. Press, 1977).

  98. 98.

    & The place of phylogeny and cladistics in evo-devo research. Int. J. Dev. Biol. 47, 479–490 (2003).

  99. 99.

    & Developmental characters in phylogenetic inference and their absolute timing information. Syst. Biol. 60, 630–644 (2011).

  100. 100.

    Phylo-evo-devo: combining phylogenetics with evolutionary developmental biology. BMC Biol. 7, 36 (2009).

Download references


We thank E. Moyroud, G. Mellers and R. Melzer for their critical reading of the manuscript and helpful comments; and E. S. Ballerini, H. D. Bradshaw, A. N. Doust, J. M. Gómez, S. A. Hodges, A. Hudson, E. Mavrodiev, G. Mellers, J. Quiles, H. Sheehan, D. E. Soltis, M. A. Streisfield and G. Theiβen for providing photographs. M.F.-M. has been supported by the Marie Curie Intra-European Fellowship LINARIA-SPECIATION (FP7-PEOPLE-2013-IEF, project reference 624396 to M.F.-M and B.J.G) and an Isaac Newton Trust Research Grant (Trinity College, Cambridge).

Author information


  1. Department of Plant Sciences, University of Cambridge, Cambridge CB2 3EA, UK.

    • Mario Fernández-Mazuecos
    •  & Beverley J. Glover


  1. Search for Mario Fernández-Mazuecos in:

  2. Search for Beverley J. Glover in:


M.F.-M. and B.J.G. wrote the manuscript jointly.

Competing interests

The authors declare no competing financial interests.

Corresponding author

Correspondence to Beverley J. Glover.