Background How the diversity of life on our planet originated is not completely understood and many questions are still open. Especially, the role of developmental robustness in evolution is an often neglected topic.
Scope Considering diverse groups of plants and animals, and employing different concepts and approaches, the authors of articles in this Special Issue try to understand better the impact of developmental robustness, phenotypic plasticity and variance on species diversity, evolution and morphological disparity.
Conclusions Several lines of theoretical considerations as well as case studies show that developmental robustness supports rather than prevents the evolution of species diversity, at least under certain circumstances. Among the possible mechanisms is the scenario that developmental robustness facilitates the synorganization of body parts, which may enable the origin of complex novelties; this then may set the ground for species radiation.
There is a huge amount of biodiversity currently present on our planet; the mechanisms by which it has been generated are only incompletely known. The authors of this article have been interested in the role of developmental processes in evolution for quite a while. Specifically, we have been reflecting on how developmental robustness affects the origin of biodiversity. Robustness of a phenotypic trait is often defined as the absence, or low level, of variation in the face of environmental or genetic perturbations ( Felix and Barkoulas, 2015 ). Intuitively, one might thus consider that robustness limits and constrains evolvability and hence biodiversity – if most environmental changes and mutations have no effect, there will be less variation for selection to act on ( Masel and Trotter, 2010 ). However, considering the highly standardized body plans of the most species-rich taxa on this planet, such as some clades of insects and flowering plants, we realized that this naïve assumption might be simplistic ( Melzer and Theißen, 2016 ). And indeed, the opposite of the view presented above – that robustness promotes evolvability – has often been claimed (e.g., Kitano, 2004 ; Wagner, 2008 ; Masel and Trotter, 2010 ).
Puzzled about the contradictions, we had intense and often controversial, but always entertaining debates with colleagues from different disciplines about the impact of developmental robustness on biodiversity. We soon realized that many of our colleagues have views that differ, often considerably, from ours. Encouraged to embark on a fruitful scientific controversy, we also became aware that clarification of matters is already hampered by problems in communication. For example, many colleagues use the same terms such as ‘robustness’ and ‘plasticity’, but with (often subtly) different meanings. Others preferred to switch terms, e.g. to use ‘canalization’ or ‘homeostasis’ rather than ‘robustness’, and ‘evolvability’ rather than ‘biodiversity’, and it was not always clear as to whether these terms eventually refer to the same phenomena, or what the differences are and why our colleagues seemed to feel more comfortable by using a different terminology.
We thought that the best way to clarify matters might be to talk things over, and so we organized a symposium entitled ‘ How does developmental robustness facilitate the evolution of biodiversity?’ at the fifth meeting of the European Society for Evolutionary Developmental Biology (EED), held in July 2014 in Vienna. Encouraged by the quality and quantity of the presentations we agreed to edit a Special Issue in Annals of Botany devoted to the topic, not least to be able to invite even more colleagues to outline their views. So here we are! Remembering a quote of one of Vienna’s most famous philosophers, Ludwig Wittgenstein (1998) – ‘ Wovon man nicht sprechen kann, darüber muss man schweigen ’ (‘Whereof one cannot speak, thereof one must be silent’) – we are very grateful that so many eminent experts did not stay silent but raised their voice loud and clear in an exciting discussion.
ROBUSTNESS: CONCEPTS, OVERVIEWS AND VIEWPOINTS
One of these voices is raised by Alessandro Minelli. By dealing with many of the critical terms in developmental and evolutionary biology, such as ‘robustness’, ‘key innovations’, ‘evolvability’ and ‘constraints’, ‘parallelism’ and ‘convergence’, ‘phenotypic plasticity’, ‘modularity’ and ‘heterochrony’, he is one of several authors that help us through the jungle of terminology ( Minelli, 2016 ; see also Abley et al. , 2016 ; Lachowiec et al. , 2016 ; Mestek Boukhibar and Barkoulas, 2016 ; Oyston et al. , 2016 ). By introducing an important pair of terms, ‘diversity’, as measured by the number of species included in a clade, and ‘disparity’, being an estimate of the lineage’s occupancy of a suitably defined n -dimensional morphospace, Minelli (2016) makes a distinction that is important in the framework of evolutionary and developmental biology. Using a number of impressive examples from animals as well as plants (highly appreciated, especially since Sandro Minelli is a zoologist by training), he demonstrates that evolutionary ‘success’ of a clade of organisms in disparity does not necessarily go together with success in diversity, and vice versa . The author hypothesizes that developmental robustness is better reflected in disparity than in diversity, but more studies on the actual patterns of diversity and disparity of a number of plant and animal taxa are required in order to test that hypothesis.
It is important to keep in mind that actual patterns of diversity and disparity are only a snapshot of a clade’s phylogenetic history. Accordingly, Oyston et al. (2016) studied disparity patterns of plant clades through evolutionary timescales. In line with Minelli (2016) , they show that disparity and diversity are essentially decoupled, with many plant lineages already possessing high disparity early during their evolutionary history ( Oyston et al. , 2016 ). The authors propose that many regions in the morphospace are occupied early during the evolution of a clade and that diversity increases by filling the intervening morphospace. Being also zoologists by education, these authors make an excellent point in suggesting that studying the disparity of plant clades deserves more attention as this could provide some fundamental insights into the similarities and differences of morphological evolution in plants versus animals.
While studying the disparity of clades through evolutionary timescales is at one end of approaches to understand morphological evolution, studying the developmental robustness of certain traits within one species is at the other end. Mestek Boukhibar and Barkoulas (2016) provide an introduction as to what robustness is and suggest experimental approaches to study it. They emphasize that it is especially important to distinguish effects on trait means from effects on trait variance when studying the genetic basis of robustness. Developmental genetics tends to be ‘mean-centric’, where screens are conducted to find mutants that deviate from the mean of a particular trait, and phenotypic averages are compared without paying too much attention to the variance. This approach has arguably been extremely successful in the past in deciphering gene regulatory circuits and their control over phenotypes. However, there are a number of studies illustrating that keeping the variance of a particular trait in a certain range can be as important as controlling the mean value of the trait ( Mestek Boukhibar and Barkoulas, 2016 ). Especially in plants, though, those studies are only just beginning to emerge.
Considering examples from animals and plants, Melzer and Theißen (2016) note that there might be a positive correlation between developmental robustness and species diversity. They thus view developmental robustness as an important but somewhat neglected factor that increases species diversity. Melzer and Theißen (2016) hypothesize that robustness at the level of the body plan of multicellular organisms may facilitate the evolution of morphological disparity ‘superimposed’ on this body plan. Considering possible evolutionary mechanisms that may explain these observations, the authors propose that developmental robustness may facilitate synorganization and subfunctionalization. Therefore, robustness may have been an important prerequisite for the evolution of more complexity, such as the evolution of new types of organs or structures, in many cases. Examples may include animals, especially insects within arthropods, and plants, for example some clades of eudicots within angiosperms. In conclusion, Melzer and Theißen (2016) suggest that beyond evolutionary novelties (or innovations) and ecological opportunities, developmental robustness is a third major driver of species diversity. Novelties, opportunities and robustness are hypothesized to constitute a self-reinforcing mechanism that may have been critical for the success of many taxa.
ROBUSTNESS VS PLASTICITY
With plasticity being the ability of organisms to alter their phenotype in response to environmental changes and robustness being the ability to maintain a specific phenotype under perturbations, both phenomena might be considered to be on opposite ends of a continuum ( Ehrenreich and Pfennig, 2016 ; Lachowiec et al. , 2016 ). Paradoxically, robustness as well as plasticity have been maintained to be drivers of phenotypic evolution and adaptive radiation ( West-Eberhard, 2003 , 2005 ; Wagner, 2008 ; Ehrenreich and Pfennig, 2016 , Melzer and Theißen, 2016) . As reviewed by Ehrenreich and Pfennig (2016) and Lachowiec et al. (2016) , plasticity is favoured when a particular trait of an organism is faced with changing environmental conditions, whereas robustness is favoured and plasticity is costly when stabilizing selection acts on a trait. When plasticity is present, new phenotypes may emerge in altered environments and selection may favour those new phenotypes and, by a process known as genetic accommodation, fix them genetically ( Ehrenreich and Pfennig, 2016 ). Along those lines of reasoning, it has been argued that genes are actually more often followers than leaders in evolutionary change ( West-Eberhard, 2003 , 2005 ; Ehrenreich and Pfennig, 2016) . This view (which is not to be confused with Lamarckian concepts of evolution, see West-Eberhard, 2003 ) places plasticity at the heart of speciation and the evolution of new forms, so where does robustness fit in this picture? One argument is that robustness allows the accumulation of cryptic genetic variation – variation that is not phenotypically expressed because of the robustness of the trait under consideration ( Masel and Siegal, 2009 ; Mestek Boukhibar and Barkoulas, 2016 ). However, when robustness breaks down, for example because of drastically changed environmental conditions, this cryptic variation is revealed, giving rise to a number of different phenotypes, some of which may be well adapted to the new environment ( Masel and Siegal, 2009 ; Ehrenreich and Pfennig, 2016 ; Mestek Boukhibar and Barkoulas, 2016 ). Beyond that, it is hypothesized that robustness also contributes to evolvability, because it increases the heritability of traits, which may subsequently contribute to an increase in morphological complexity and to the origin of evolutionary innovations and diversifying selection (see above) ( Peterson et al. , 2009 ; Melzer and Theißen, 2016 ).
It also remains to be noted that stochastic and programmed plasticity can be distinguished ( Lachowiec et al. , 2016 ; see also Abley et al. , 2016 ). Whereas programmed plasticity refers to predictable phenotypic changes in response to environmental changes, stochastic plasticity is observed when the phenotypic changes are unpredictable ( Lachowiec et al. , 2016 ). For the evolution of morphological complexity as hypothesized above, programmed plasticity as well as robustness might provide a suitable starting point, whereas stochastic and programmed plasticity both may contribute to phenotypic evolution through genetic accommodation. The contribution of Abley et al. (2016) also deals with the evolutionary importance of phenotypic plasticity, variability and developmental robustness. These authors argue that plastic plant phenotypes which vary between different environments or variable phenotypes that vary stochastically within an environment could be of advantage under some circumstances. They distinguish between invariant, plastic and variable phenotypes, but they see no opposition between robustness on one hand and plasticity and variability on the other hand. Abley et al. (2016) rather maintain that for some plant traits, mechanisms may have evolved to generate variability or plasticity in a reliable, which means robust, manner.
SAY IT WITH FLOWERS: CASE STUDIES ON FLORAL ARCHITECTURE
Flowers provide a rich source of material for investigations on the evolutionary importance of developmental robustness and phenotypic plasticity, and it shows. Both Apocynaceae (a family of eudicots) and Orchidaceae (a family of monocots) produce pollinia and pollinaria, but no other angiosperms do so. Since the two groups of plants are only very distantly related within angiosperms, these traits obviously originated by convergent evolution. Endress (2016) presents a detailed comparative study of the flowers of these plants which highlights the developmental and evolutionary preconditions for these traits and a number of other convergences. According to Endress (2016) , these other convergences are largely a precondition for, or a result of, synorganization of the parts of the flower. With about 25 000 species, Orchidaceae are one of the most diversified families of flowering plants. Most interestingly, species and genus richness are especially concentrated in those subclades with the highest flower synorganization (Orchidoideae and Epidendroideae with about 21 500 species). Apocynaceae have almost 5000 species. Also in this case, the clades with the most synorganized flowers (Asclepiadoideae and Secamonoideae) have more species (3180) than all other subclades of the family taken together ( Endress, 2016 ). Even though different habitats and different ages of the families make a direct comparison difficult, it is tempting to hypothesize that the synorganization of the flower played an important role in the diversification in both families of angiosperms ( Endress, 2016 ). We propose that developmental robustness is always of special importance if two or more structures need to function in a highly integrated manner with each other, as in case of synorganization ( Melzer and Theißen, 2016 ). Thus if synorganization requires robustness, Apocynaceae and Orchidaceae support the view that developmental robustness furthers species diversity via synorganization.
Rutishauser (2016) outlines a somewhat different scenario. He reports his long-standing botanical interest: angiosperms that can be considered being ‘morphological misfits’, because they appear as large morphological deviations from the norm (the typical root–shoot body plan) by revealing blurred (‘fuzzy’) organ identity. Intriguing examples are provided by the Podostemaceae (river-weeds) and Utricularia (bladderworts, Lentibulariaceae). Bladderworts are carnivorous plants that develop sucking traps and live as submerged aquatics, humid terrestrials or as epiphytes. Most river-weeds live as submerged haptophytes attached to rocks in tropical river-rapids and waterfalls. Interestingly, despite their fuzzy vegetative morphology, both river-weeds and bladderworts have stable flower baupläne , indicating that ‘release’ or ‘decanalization’ of a body plan can affect only part of the ontogeny, while other parts remain highly constrained. Thus river-weeds and bladderworts are vegetative ‘misfits’ but floral conformists. Rutishauser (2016) points out that in case of bladderworts and river-weeds, the release from a classical root-shoot bauplan facilitated rather than prevented the evolution of species diversity. Even though both clades are only of moderate size (about 310 and 230 species in Podostemaceae and Utricularia , respectively), one may conclude that the ecological circumstances may decide as to whether developmental robustness has a positive or negative effect on species diversity.
Petal number variation is an emerging topic in studies on variance, plasticity and robustness. It is also covered by several articles in this Special Issue. Whereas most Brassicaceae have four petals, in Cardamine hirsuta the petal number varies stochastically. When studying the genetic basis of this variation, Monniaux et al. (2016) identified four quantitative trait loci (QTLs). Importantly, the authors distinguish between average petal number and its variation and show that both traits may have a common genetic basis. Thus, the four QTLs identified may not represent robustness genes in the strict sense (i.e. only affecting trait variation, not trait mean); they do, however, provide an excellent starting point for further analyses of petal number variation in flowering plants, a trait that appears to be extremely robust in many core eudicot and monocot lineages but much more relaxed in other angiosperms, especially early diverging lineages ( Fig. 1 ) ( Warner et al. , 2008 , 2009 ; Endress, 2011 ; Becker, 2016 ; Melzer and Theißen, 2016 ).
As already mentioned above, in order to better understand robustness and plasticity it is important not only to consider the mean of traits but also their variance ( Mestek Boukhibar and Barkoulas, 2016 ). Kitazawa and Fujimoto (2016) provide a detailed study that does just that, using the intraspecific variation of the organ numbers within flowers of Ranunculaceae, and the numbers of flowers within the capitula of Asteraceae as examples. Employing four elementary statistical quantities the authors assess not only the robustness of the morphologies but also how flower structures diversify during evolution.
Beyond perianth organ number, the shape of the corolla and the degree of floral integration are certainly of considerable importance for many plant–pollinator interactions, particularly for flowers with specialized pollination systems. It might therefore be expected that specialized flowers also possess a relatively high degree of developmental robustness in corolla shape and floral integration, even though experimental evidence supporting this assumption is still relatively scarce ( Rosas-Guerrero et al. , 2011 ; Armbruster, 2014 ). The question is, however, to what extent robustness in corolla shape is also present in taxa with a more generalized pollination system like those found in Brassicaceae. Surprisingly, Gómez et al. (2016) found that corolla shape is related to pollination niches in Brassicaceae and that corolla shape may evolve in response to selection by generalist pollinators (see also Gómez et al. , 2015 ). Moreover, depending on the pollination niche, the magnitude of corolla shape variation differed, indicating that certain pollination niches are associated with a lower degree of developmental robustness than others ( Gómez et al. , 2016 ).
Strelin et al. (2016) investigated flowers of the genus Caiophora (Laosideae) to learn whether developmental robustness to genetic and non-genetic perturbations limits the phenotypic space available for future evolutionary changes. There is evidence that flower diversification into different pollination syndromes in Caiophora took place during a recent adaptive radiation in the central Andes mountain region. This process involved transitions from bee to hummingbird and small rodent pollination. Strelin et al. (2016) examined whether the adaptive radiation of pollination syndromes in Caiophora occurred through ontogenetic scaling or involved a departure from the ontogenetic pattern basal to this genus. By demonstrating that variation in the size and shape of Caiophora flowers does not overlap with the pattern of ontogenetic allometry of early diverging species the authors reveal that the adaptive radiation of Caiophora involved significant changes in the developmental pattern of the flowers, thus rejecting the ontogenetic scaling hypothesis.
Floral diversification and speciation is also the topic of a contribution by Ronse De Craene and Bull-Hereñu (2016) . Obdiplostemonous flowers have two sets of stamens in alternating whorls, with the outer whorl stamens opposite to the petals. Ronse De Craene and Bull-Hereñu (2016) review the impact of obdiplostemony on floral diversification and species evolution. They argue that even though obdiplostemony is only of transient taxonomic significance, it illustrates how developmental flexibility is responsible for highly different floral morphs.
THE ROLE OF GENES AND REGULATORY NETWORKS
If developmental robustness is advantageous, at least under many evolutionary conditions, one may expect that there are genes and gene regulatory networks that confer it. Indeed, ‘canalizing genes’ that ‘confer robustness’ have been described in the literature, even though it often remains unclear (and is generally difficult to determine) whether these genes were selected during evolution for this specific function (reviewed by Felix and Barkoulas, 2015 ). An interesting candidate might be SUPERMAN ( SUP ), a cadastral gene that controls the boundaries of sexual organs in the flowers of Arabidopsis thaliana . Breuil-Broyer et al. (2016) characterized an impressive allelic series of the SUP gene in quite some detail. Their results support the view that SUP functions have been important for transforming male/female gradients into sharp male and female whorl and organ identities, and for increasing developmental robustness (‘homeostasis’ in the authors’ words) of the flower during evolution. Breuil-Broyer et al. (2016) suggest that this may have been achieved by the incorporation of a system that patterns the meristem of the floral axis into the female (carpel) developmental program.
Switching gears from the analysis of individual genes to gene regulatory networks (GRN), Becker (2016) reviews the role of floral homeotic genes on the developmental robustness and phenotypic plasticity of flowers in the Ranunculales. Flowers from this taxon often show much more variance in key traits than flowers from core eudicots. Becker (2016) advertises Ranunculales as a promising model system to investigate the genetic basis of robustness and plasticity in a framework of ecological evolutionary developmental biology (eco-evo-devo).
On a more general basis, hourglass-type gene regulatory networks have been suggested to confer robustness ( Kitano, 2004 ). As reviewed by Cridge et al. (2016) , there is evidence that animal development follows such an hourglass pattern, and that plants as well as animals possess a transcriptomic hourglass with an especially conserved gene expression profile during the ‘middle’ stage of embryogenesis ( Domazet-Loso and Tautz, 2010 ; Kalinka et al. , 2010 ; Quint et al. , 2012 ; Cridge et al. , 2016 ). Cridge et al. (2016) speculate that the hourglass structure of the transcriptomic network in plant and animal embryogenesis provides robustness at one critical stage during development and constrains the morphological space the organism is able to occupy, but at the same time allows for morphological diversification within those limits.
The relevance of plasticity and robustness for the origin of evolutionary innovations and species diversity is still far from being understood and some of the arguments are indeed heavily debated ( Laland et al. , 2014 , Wray et al. , 2014 ). As shown by the papers of this Special Issue, some of the most important questions are: Why are some body plans evolutionarily more conserved than others? Is that conservation always adaptive? How are the developmental concepts of robustness, plasticity and constraints related to the diversity and disparity of entire clades? What are the molecular mechanisms conferring robustness? We are far from having conclusive answers, but the papers presented in this Special Issue excellently set the stage for further studies.
We imagine that having us as guest editors for this Special Issue required quite some robustness as well as plasticity from the entire Annals of Botany team, and we are very grateful for that. Very special thanks to Trude Schwarzacher, Pat Heslop-Harrison and David Frost for being the main evolutionary driving forces during the preparation of this Special Issue. Many thanks also to Susanne Schilling for providing the images for the figure. We would also like to thank all the authors and referees who gave their time to write and review manuscripts, respectively, and without whose enthusiasms and help we could not have achieved anything.