Developmental mechanisms facilitating the evolution of bills and quills

Abstract

Beaks and feathers epitomize inimitable avian traits. Within individuals and across species there exists astounding diversity in the size, shape, arrangement, and colour of beaks and feathers in association with various functional adaptations. What has enabled the concomitantly divergent evolution of beaks and feathers? The common denominator may lie in their developmental programmes. As revealed through recent transplant experiments using quail and duck embryos, the developmental programme for each structure utilizes mesenchyme as a dominant source of species-specific patterning information, acts as a module of closely coupled molecular and histogenic events, and operates with a high degree of spatial and temporal plasticity. By synergizing these three features, the developmental programmes underlying beaks and feathers likely have the essential potential to react spontaneously to novel conditions and new gene functions, and as a consequence are well equipped to generate and accommodate innovative phenotypes during the course of evolution.

The ousel cock so black of hue,

With orange-tawny bill,

The throstle with his note so true,

The wren with little quill …

(William Shakespeare, A Midsummer Night's Dream)

Introduction

The bills and quills of birds represent some of the most diversified and highly adapted anatomical structures throughout vertebrates. Their many physical attributes can change rapidly and often reflect ecological and functional demands with astonishing precision. These properties were essential for formulating early theories of evolution via natural selection, as famously exemplified by the beaks of Galapagos Finches that Charles Darwin originally described (Darwin, 1859) or by the feathers of Malayan Birds of Paradise that Alfred Wallace first characterized (Wallace, 1869). Ever since Darwin and Wallace, much attention has been focused on the role of external factors such as the environment or mating behaviour in directing the course of beak and feather evolution. By contrast, modest consideration has been given to cellular and molecular mechanisms that generate species-specific differences. But recent studies on this topic are beginning to provide a deeper understanding of how beaks and feathers become patterned, and more significantly, they are leading to a fundamental realization that such developmental programmes may also serve as internal factors for facilitating swift and widespread morphological change during avian evolution.

Beaks and feathers share at least three features in their developmental programmes that potentially predispose them to expeditious evolutionary transformations. First, their developmental programmes each employ epithelial and mesenchymal cells to produce composite structures, but the mesenchyme seems to act as the dominant source of species-specific patterning information. In this capacity, mesenchyme may serve as a conductive target of natural selection and a driving force for generating phenotypic variation. Second, developmental programmes for beaks and feathers operate as modules of tightly integrated molecular and histogenic events. These modules involve series of reciprocal epithelial–mesenchymal interactions, which are mediated by many of the same signalling pathways and are regulated by equivalent cellular mechanisms. By enabling developmental programmes to be enacted in a variety of new contexts, modularity may be an essential element for enhancing the evolvability of anatomical complexes such as beaks and feathers. Third, developmental programmes for beaks and feathers possess an apparent capacity to fluctuate in space and time. This feature is a hallmark of plasticity, which is a measure of the extent to which ontogenetic systems can respond to internal and external perturbations and produce an integrated and sustainable phenotype. While natural selection can enable a given phenotype to become cleverly fitted to its environment, the key to achieving this state is adaptability, and adaptability may arise as a function of the degrees of spatiotemporal plasticity inherent in a developmental system.

The role of epithelia vs. mesenchyme during beak and feather morphogenesis

Epithelial and mesenchymal cells have distinctly different embryonic origins and unique patterning abilities during morphogenesis. The epithelial components of beaks and feathers are derived from surface ectoderm and are arranged as polarized sheets of tightly connected cells. Surface ectoderm gives rise to the epidermis, which is a stratified tissue organized into multiple layers (Romanoff, 1960; Hamilton, 1965; Yasui & Hayashi, 1967). The upper layer of the mature epidermis is the non-living stratum corneum, which characteristically forms cornified (i.e. keratinized) tissues. In birds, the epidermis contributes to feathers, scales, beaks and egg teeth (Kingsbury et al. 1953; Lucas & Stettenheim, 1972; Sawyer et al. 1984; Couly & Douarin, 1988; Pera et al. 1999; Yu et al. 2004). A thin basement membrane divides epidermis from underlying dermis, which is derived from mesenchyme.

Mesenchyme consists of loosely associated stellate-shaped cells, which in the trunk and caudal regions of the head arise from mesoderm, and in the face and portions of the neck predominantly come from the cranial neural crest (Noden, 1978, 1986; Couly et al. 1992; Olivera-Martinez et al. 2000, 2004a; Matsuoka et al. 2005). Neural crest mesenchyme originates along the dorsal surface of the neural tube during neurulation and migrates throughout the craniofacial complex (Tosney, 1982; Hall & Hörstadius, 1988). Neural crest cells also make pigment-producing melanocytes, which infiltrate the epidermis and are the source of colour in beaks and feathers (Rawles, 1948; Cramer, 1991; Le Douarin & Dupin, 1993; Bronner-Fraser, 1994; Hirobe, 1995).

Although beaks vary greatly across groups of birds, at early embryonic stages they all arise from comparable primordia, tissues and cells. The upper aspect of the beak is derived from the frontonasal and paired maxillary primordia, whereas the lower portion forms from paired mandibular primordia (Fig. 1). Cranial neural crest cells are the principal source of mesenchyme for the facial primordia. During beak development, neural crest cells generate skeletal and connective tissues, mesoderm gives rise to vascular endothelium and voluntary muscles, nasal placodes produce olfactory epithelium, pharyngeal endoderm lines part of the oral cavity, and surface ectoderm forms superficial cornified layers (Lièvre & Douarin, 1975; Noden, 1978, 1983; Couly & Douarin, 1988; Couly et al. 1992, 1993, 1995; Webb & Noden, 1993). Much current experimental work is steadily illuminating the precise roles that these tissues play in patterning the beak. By and large, epithelia that surround the developing facial primordia seem to supply positional cues and maintenance factors necessary for patterned outgrowth. For example, signalling by epithelium along the frontonasal primordium is required for proper expansion and orientation of skeletal elements along the proximodistal, mediolateral and dorsoventral axes (Hu & Helms, 1999; Hu et al. 2003). Similarly, endodermal epithelium of the pharynx is necessary for the formation and arrangement of bone and cartilage in the facial skeleton (Couly et al. 2002). Thus, epithelia of both ectodermal and endodermal origin that surround the facial primordia serve as local sources of morphogenetic signals, which elicit programmatic responses from underlying mesenchyme (Richman & Tickle, 1989; Langille & Hall, 1993; Mitsiadis et al. 2003; Santagati & Rijli, 2003; Le Douarin et al. 2004).

Fig. 1.

Morphogenesis of the avian beak. (A) At embryonic stage (HH) 9.5 (Hamburger & Hamilton, 1951), neural crest cells (dorsal view) migrate from the caudal forebrain (fb) and rostral midbrain (mb) and produce frontonasal structures (pink). Cells from the caudal midbrain and first two rhombomeres (r) of the hindbrain give rise to maxillary and mandibular derivatives (light blue). The hyoid arch forms from r4 (green) and the third arch forms from r6 (brown). Neural crest cells migrate through and around mesodermal mesenchyme (m; grey). (B) By HH25, the frontonasal (fn), maxillary (mx), mandibular (ma) and hyoid (hy) primordia (sagittal view) are surrounded by surface ectoderm (se; light purple), pharyngeal endoderm (pe; yellow) and forebrain neuroepithelium (fb; dark blue), and contain contributions from the neural crest, nasal placode (np; dark blue) and cranial ganglia (V, VII, IX; dark blue). Mesoderm (m) produces muscles, vascular endothelial cells, and some skeletal tissues. (C) By HH40, neural crest cells produce cartilaginous and bony elements in the facial and jaw skeletons (pink, light blue, green and brown) whereas mesoderm forms much of the caudal cranial vault (grey) and basicranium caudal to the pituitary. Based on data and drawings from a variety of sources (Noden, 1978, 1988; Couly et al. 1993; Köntges & Lumsden, 1996; Schneider, 1999; Schneider & Helms, 2003). (D,E) At HH25, the frontonasal (fn), maxillary (mx) and mandibular (ma) primordia of quail and duck appear equivalent in proportion and shape but not size (frontal view). Scale bars = 1 mm.

A recent study investigated the patterning potential of cranial neural crest mesenchyme during beak morphogenesis (Schneider & Helms, 2003). Neural crest cells destined to form the beak were transplanted from quail to duck generating chimeric ‘quck’ embryos, and from duck to quail creating chimeric ‘duail’ embryos (Fig. 2). Such an experimental design exploits three features that distinguish these species of birds. First, quail beaks are short, narrow and convex in comparison with duck bills, which are long, broad and flat. This provides a simple way to determine whether or not facial structures that result from the transplants more closely resemble the donor or host. Second, quail and duck embryos have significantly different maturation rates. If donor cells maintain their own developmental timetable in the host environment, then the extent to which neural crest regulates other tissues involved in beak patterning can be readily assessed based on whether morphogenesis is accelerated or delayed in chimeras. Third, quail cells can be detected by using a ubiquitous nuclear marker not present in duck. This allows donor- and host-derived structures to be identified unequivocally.

Fig. 2.

The cellular origins of species-specific beak morphology. (A,B) The beaks of quail embryos are short and blunt whereas those of duck are long and broad. (C) Transplants of cranial neural crest cells, which are destined to form the beak, from quail donors to duck hosts produce chimeric ‘quck’ embryos with quail-like beak morphology (Schneider & Helms, 2003). Note that the quail-like quck has webbed feet (arrow), which is indicative of the duck host.

Quail–duck transplants reveal that neural crest cells provide species-specific information for patterning the beak. When transplanted into duck embryos, quail neural crest cells gave rise to beaks like those found in quail. Reciprocal transplantations of duck presumptive upper bill neural crest into quail hosts produced analogous alterations. The bills of chimeric duail resembled those of duck in shape and relative length, and were composed of elements derived mainly from duck donor neural crest. In these quail–duck experiments, donor neural crest cells also transformed host tissues. For example, duck have an egg tooth that is a flat epidermal nail at the tip of the bill, whereas quail posses an egg tooth that is a conical protuberance of hard keratin (Lucas & Stettenheim, 1972). The quck egg tooth, despite being formed from non-transplanted host tissue, appeared like that found in quail, whereas the duail egg tooth resembled that of the duck. Additionally, quck beak epidermis was thick as in quail instead of being lithe as in duck. Thus, neural crest cells guided their own morphogenesis and they also patterned non-neural crest-derived beak tissues.

Similar results were obtained in a separate series of experiments in which neural crest cells fated to form skeletal elements around the jaw joint were transplanted between quail and duck embryos (Tucker & Lumsden, 2004). As in the case of beak morphology, the shape of cartilages such as the retroarticular process and entoglossum, which differ substantially between quail and duck in association with their unique feeding behaviours, formed according to the identity of the donor species. Overall, both investigations using quail–duck chimeras demonstrate the essential regulatory role played by the mesenchyme in establishing species-specific morphology of the beak and jaw apparatus. Such findings build upon classic transplant experiments using salamanders and frogs that were among the earliest to suggest a role for neural crest cells in patterning the jaw skeleton (Andres, 1949; Wagner, 1959).

Like beaks, feathers form as composite structures derived from mesenchyme and epithelia (Fig. 3). Histologically, feather formation begins with the aggregation of mesenchyme into a thin, homogeneous layer of dense dermis under the epithelium (Wessells, 1965; Brotman, 1977; Mayerson & Fallon, 1985). The epithelium then thickens into a localized epidermal placode, the mesenchyme aggregates into a dermal condensation, the placode and mesenchyme rise above the surface of the integument, and both tissues proliferate and differentiate (Pispa & Thesleff, 2003; Olivera-Martinez et al. 2004b). Morphologically, feathers form as distinct buds in continuous rows that make up tracts (Lucas & Stettenheim, 1972). Each row is added sequentially, and as a result, tracts contain feather buds at successive stages of development.

Fig. 3.

The origins of species-specific feather pattern. (A) Cranial feather buds form through interactions between neural crest-derived dermis and overlying epidermis, which would be derived from donor and host, respectively. In both quail and duck at HH33, there is little histological evidence for cranial feather development, but by HH34, epithelial placodes form in the epidermis and the mesenchyme aggregates into dense dermis. By HH36, feather buds contain dermal condensations and they begin to rise above the surface of the integument. Long buds form after HH37. (B,C) Quail cranial feather buds are large and widely spaced in comparison with those of duck, which are smaller and spaced closer together, as shown schematically. (D) Quail and duck embryos stage-matched for surgery at HH9.5 subsequently diverge in stage due to their different rates of maturation. Cranial feather placodes arise in duck almost 3 days after they are observed in quail (arrows). Modified from Eames & Schneider (2005). (E) In chimeric quck, cranial feather tracts patterned by quail donor mesenchyme (left side) adjoin those patterned by duck host mesenchyme (right side). Note that the quail-like feathers are at the long bud stage whereas those derived from the duck host linger as placodes.

To explore the function of mesenchyme during cranial feather morphogenesis, a recent study also exploited the quail–duck chimeric system (Eames & Schneider, 2005). The embryonic feather buds of Japanese quail are relatively large, widely spaced and pigmented whereas those of the white Pekin duck are smaller, closely arranged and un-pigmented. Again, premigratory cranial neural crest cells were exchanged between quail and duck embryos during the time of neurulation. In both types of resulting chimeras, cranial feather morphology developed in accordance with the identity of the donor species. Chimeric quck had areas of black and brown quail-like feathers clustered among short white duck host feather buds. Reciprocal transplants of cranial neural crest cells from duck donors into quail hosts produced chimeric duail embryos with duck-like feather patterns. These transplants support the essential function of the dermis as a primary source of patterning information in the feather tracts. This finding is consistent with other recombination experiments performed in the trunk at later embryonic stages such as those using chick and duck tissues, which suggested that the dermis transmits species-specific information for feather morphology (Dhouailly, 1967, 1970) as well as instructions for the time of appearance, location, size, number and morphological identity of feathers (Cairns & Saunders, 1954; Saunders & Gasseling, 1957; Rawles, 1963; Wessells, 1965; Linsenmayer, 1972; Dhouailly, 1973; Dhouailly & Sawyer, 1984; Song & Sawyer, 1996; Fliniaux et al. 2004; Prin & Dhouailly, 2004).

Developmental modules as facilitators of beak and feather evolution

Another significant outcome of using a quail–duck chimeric system was the discovery that developmental programmes underlying beaks and feathers behave as modules of integrated histogenic and molecular events, which are synchronized through epithelial–mesenchymal signalling interactions. By design, these recent quail–duck transplant experiments exchanged premigratory neural crest cells between embryos matched at the same stage of neurulation so that donor mesenchyme and host epithelium could experience continuous interactions with one another starting as close to their genesis as possible. Moreover, owing to intrinsic species-specific differences in growth rates (17 days to hatching for quail vs. 28 days for duck), chimeric quck contain quail donor mesenchyme, which ordinarily would follow a relatively faster developmental timetable, adjacent to duck host epithelia, which normally would undergo a relatively slower embryogenesis. By contrast, chimeric duail have relatively delayed duck donor mesenchyme juxtaposed with quail host epithelia that usually would develop more rapidly. Consequently, host epithelium is challenged to respond to progressively asynchronous signalling by donor neural crest-derived mesenchyme. In the case of quck feather development, donor mesenchyme initiated morphogenesis three stages earlier than normal, while in the duail, feather development was delayed three stages (Eames & Schneider, 2005). In absolute time for the stages examined, this translates to roughly 4 days for duck and 2½ days for quail, which for each species represents almost 15% of their total incubation period. Thus, these experiments reveal that neural crest cells not only determine spatial pattern, but also even more significantly, establish when host structures form, probably by governing the timing of signalling interactions between the mesenchyme and overlying epithelium.

Precisely timed epithelial–mesenchymal signalling interactions are required for the formation of many vertebrate organ systems, including the limbs, facial primordia, teeth and integument (Saunders & Gasseling, 1968; Tonegawa, 1973; Salaun et al. 1986; Fisher, 1987; Wedden, 1987; Lumsden, 1988; Sharpe & Ferguson, 1988; Richman & Tickle, 1992; Francis-West et al. 1998; Mitsiadis et al. 1998; Schneider et al. 1999, 2001; Shigetani et al. 2000; Hu et al. 2003; Pispa & Thesleff, 2003). At each stage during their interactions, the epithelium and mesenchyme may function in different capacities, and act either instructively or permissively in a manner that allows morphogenesis to advance. To understand the specific role of the mesenchyme or epithelium at any given point in time, tissues typically have been separated and then recombined in vitro. For example, recombinations of heterochronic dermis and epidermis from wild-type chicks and featherless mutants indicate that, initially, the dermis can induce epidermal placodes, but this ability disappears without proper feedback from the epidermis (Viallet et al. 1998). At later stages, the epithelium can act instructively by directing where teeth form (Thesleff & Sharpe, 1997; Tucker et al. 1998; Wang et al. 1998), by determining whether epidermis differentiates into feathers or scales (Widelitz et al. 2000; Prin & Dhouailly, 2004) and by specifying feather branching patterns (Harris et al. 2002; Yu et al. 2002).

In the case of quck beaks, mesenchyme is able to bring about species-specific morphology by dominating the initial interactions with overlying epithelium, and in particular by regulating the mesenchymal and epithelial expression of genes known to affect facial patterning (Schneider & Helms, 2003). Together, transcription factors and secreted molecules exhibited temporal shifts in the initiation of their expression consistent with differences in maturation rates between donor and host cells, providing evidence that quail neural crest cells created quail beaks on ducks by maintaining their own molecular programmes and by altering patterns of gene expression in non-neural crest host tissues.

These findings were further substantiated by a more systematic analysis of donor-induced changes to gene expression in the context of feather morphogenesis. To determine the extent to which mesenchyme directs feather morphogenesis on the molecular level, control quail, duck and chimeric embryos were compared at successive stages using in situ hybridization to detect mRNA transcripts. The temporal expression patterns of members and targets of the Bone Morphogenetic Protein (BMP), Sonic Hedgehog (SHH) and Delta/Notch signalling pathways were analysed as these have been shown to inhibit and promote feather development (Ting-Berreth & Chuong, 1996; Widelitz et al. 1997; Crowe et al. 1998; Morgan et al. 1998; Patel et al. 1999; Chuong et al. 2000; 2001; Ashique et al. 2002; Yu et al. 2002; Pispa & Thesleff, 2003). For each signalling pathway examined in chimeras there was a significant change in the timing of expression in both the mesenchyme and the epithelium consistent with the embryonic stage of the donor neural crest. Thus, mesenchyme regulates signalling by the BMP, SHH and Delta/Notch pathways and functions as an overriding source of spatial and temporal patterning information during cranial feather morphogenesis. These findings corroborate other experimental data demonstrating the extent to which mesenchymally mediated signalling affects development of both feathers and beaks. For example, bmp4, which is expressed in presumptive feather dermis from initial stages of mesenchymal aggregation onwards, initiates and patterns feather tracts (Jung et al. 1998; Noramly & Morgan, 1998; Scaal et al. 2002). Similarly, during beak morphogenesis, differential domains of bmp4 expression correspond to species-specific variations in the size and shape of the facial primordia among chicks, ducks and Darwin's finches (Abzhanov et al. 2004; Wu et al. 2004).

As has been elucidated through the quail–duck chimeric system, mesenchyme clearly plays an essential role in establishing where and when molecular and histogenic events begin during beak and feather development. Once these developmental programmes are initiated, their entire internal sequences seem to play out automatically due to the series of inductive reciprocal interactions stemming from the mesenchyme to adjacent epithelia. Because each molecular and histological event during beak and feather morphogenesis appears to be mechanistically coupled to the next, certain spatiotemporal modifications to the mesenchyme during development may be more likely to cause quick and saltatory changes during evolution. Such phenomena have long been associated with heterochrony, which entails alterations to the timing and rates of ontogenetic events (de Beer, 1930; Gould, 1977, 1982; Alberch et al. 1979; Raff & Kaufman, 1983; Hall, 1984; Alberch & Blanco, 1996; Smith, 2002, 2003), and are fundamental to modern concepts of modularity, in which developmental programmes act as relatively autonomous units for evolutionary diversification (Raff, 1996; Bolker, 2000; West-Eberhard, 2003; Schlosser & Wagner, 2004). Modules are self-directing hierarchical entities that can be persistently iterated during development due to the inductive relationships among their constituent parts, and in this capacity they have been viewed as a key mechanism for feather evolution (Prum, 1999; Harris et al. 2002; Prum & Brush, 2002; Prum & Dyck, 2003). Moreover, modularity seems to be a common feature of various developmental programmes operating during craniofacial morphogenesis, which has been made most apparent by those striking mutant phenotypes where the morphological identities of the facial primordia and more caudal pharyngeal arches are transformed homeotically (Rijli et al. 1998; Smith & Schneider, 1998; Grammatopoulos et al. 2000). For example, simultaneously inhibiting BMP signalling and augmenting the amount of available retinoic acid in chick embryos can turn the maxillary primordium into a frontonasal process (Lee et al. 2001). Likewise, knocking out certain homeobox transcription factors, which are predominantly expressed in neural crest-derived mesenchyme, can convert lower jaws into upper jaws in mice (Depew et al. 2002). Such gain- and loss-of-function experiments reveal the extent to which interreliant gene networks, especially those that mediate epithelial–mesenchymal signalling interactions, can be modularly activated and re-deployed during development, and as a consequence trigger dramatic anatomical transformations.

Developmental plasticity as a means to enhance the adaptability of beaks and feathers

An equally potent mechanism promoting the diversification of bird beaks and feathers appears to be the preservation of spatiotemporal plasticity during development. Plasticity embodies the natural ability of some ontogenetic systems to encounter intrinsic and extrinsic perturbations yet still construct cohesive and viable phenotypes (Raff & Kaufman, 1983; West-Eberhard, 2003; Schlosser & Wagner, 2004). By its very essence, plasticity can enhance the adaptability of developmental programmes and should be favoured under natural selection (West-Eberhard, 1989). In this context, plasticity appears to be a defining characteristic of many developmental programmes that utilize mesenchyme, especially cranial neural crest-derived mesenchyme (Le Douarin et al. 2004), which has almost certainly enhanced the evolutionary potential (i.e. adaptability) of numerous morphological complexes such as the pharyngeal and rostral portions of the vertebrate head (Gans & Northcutt, 1983), as well as the lateral wall of the mammalian skull (Smith & Schneider, 1998; Schneider, 1999).

Quail–duck transplants reveal that plasticity is an elemental aspect of the developmental programmes underlying bird beaks and feathers. In response to donor neural crest mesenchyme, host tissues somewhat seamlessly incorporate morphogenetic modifications introduced by internal stimuli, which in the case of beaks and feathers include species-specific changes to molecular and histogenic events. Conversely, neural crest mesenchyme freely implements its programme for developmental rates, timing and spatiality, and subsequently fabricates a donor phenotype in hosts. In effect, one genotype partially overrides that of the other and for the most part such perturbations are accepted with acquiescent agility, which can be viewed as an emergent property of hierarchical modularity interacting with spatiotemporal plasticity. This observation is not meant to minimize the fact that some structural malformations do arise seemingly as a consequence of miscommunications either between components straddling an interface of donor and host mesenchyme, as is the case for clefting defects often observed along the epithelial seams of chimeric facial primordia, or across spatially constrained systems where the chimeric embryo struggles to assimilate species-specific disparities in size, which seems to be the story for the eye and its neural crest-derived supporting tissues. But overall, the nature of their developmental programmes not only allows quail–duck chimeras to ‘adapt’ to experimentally induced changes, but also probably enables beak and feather phenotypes to evolve rapidly in response to varying environmental conditions, functional requirements or sexual selection.

Conclusion

The idea that development, or ontogeny, plays a central role in shaping the course of evolution, or phylogeny, has a long and colourful history. Embryological work during the early part of the twentieth century was pivotal in advancing the notion that ontogenesis can be a source of innovation and change rather than a subservient artefact of phylogenesis (Garstang, 1922; de Beer, 1930). Much subsequent research has also emphasized how development can generate and mediate phenotypic diversity (Alberch, 1980; 1982; Maderson et al. 1982; Maynard-Smith et al. 1985; Wake & Roth, 1989; Northcutt, 1990; Raff, 1994; Gilbert et al. 1996). Bird beaks and feathers, which have clearly achieved extraordinary variation, offer an exemplary case study for understanding the way ontogeny can regulate phylogeny. Developmental programmes underlying beaks and feathers consist of mesenchymal progenitors that serve as a primary source of species-specific patterning information, molecular and histogenic events that unfold as integrated modules, and morphogenetic systems characterized by spatiotemporal plasticity. Given these prevailing properties, small programmatic changes to mesenchyme during development might be sufficient to provide the material basis for evolution. Furthermore, with mesenchyme as a target, natural selection has explicit means for modifying the phenotypes of multifarious morphological complexes such as the bills and quills of birds.