Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Jul 11.
Published in final edited form as: Evolution. 2020 Jul 11;74(9):2121–2133. doi: 10.1111/evo.14051

Variations of Mesozoic feathers: Insights from the morphogenesis of extant feather rachises

Shuo Wang 1,2, Wei-Ling Chang 3, Qiyue Zhang 4, Menglu Ma 5, Feng Yang 6, De Zhuo 7, Harn I-Chen Hans 1,8, Rui Yang 5, Ping Wu 1, Michael Habib 1,9, Wen-Tau Juan 3,10, Cheng-Ming Chuong 1,11
PMCID: PMC13349713  NIHMSID: NIHMS1964876  PMID: 32614075

Abstract

The rachises of extant feathers, composed of dense cortex and spongy internal medulla, are flexible and light, yet stiff enough to withstand the load required for flight, among other functions. Incomplete knowledge of early feathers prevents a full understanding of how cylindrical rachises have evolved. Bizarre feathers with unusually wide and flattened rachises, known as “rachis-dominated feathers” (RDFs), have been observed in fossil nonavian and avian theropods. Newly discovered RDFs embedded in early Late Cretaceous Burmese ambers (about 99 million year ago) suggest the unusually wide and flattened rachises mainly consist of a dorsal cortex, lacking a medulla and a ventral cortex. Coupled with findings on extant feather morphogenesis, known fossil RDFs were categorized into three morphotypes based on their rachidial configurations. For each morphotype, potential developmental scenarios were depicted by referring to the rachidial development in chickens, and relative stiffness of each morphotype was estimated through functional simulations. The results suggest rachises of RDFs are developmentally equivalent to a variety of immature stages of cylindrical rachises. Similar rachidial morphotypes documented in extant penguins suggest that the RDFs are not unique to Mesozoic theropods, although they are likely to have evolved independently in extant penguins.

Keywords: contour feather, Burmese amber, developmental constraint, evo-devo, Late Cretaceous

Feathers are one of the most complicated integumentary derivatives, encompassing hierarchical branches of rachis, barbs, and barbules, in addition to the proximal calamus (Lucas and Stettenheim 1972; Prum and Dyck 2003). Feathers are either radially symmetric, bilaterally symmetric, or asymmetric (Lucas and Stettenheim 1972; Xu and Guo 2009), depending on whether a rachis is present and whether it is centrally localized in the feather vane (Lucas and Stettenheim 1972; Foth 2011). The widely accepted developmental biology model of feathers suggests that the cylindrical rachis is formed through the fusion of barb ridges (Prum and Dyck 2003; Yue et al. 2006) and consists of a dense outer cortex and a spongy central medulla. This structure makes feathers light, flexible, and strong enough to withstand large aerodynamic forces encountered during flight (Lucas and Stettenheim 1972). A recent study revealed that the extant feather properties, including mechanical strength, are primarily determined by several factors including the cross geometry of the rachises, cell shape and bulk size of the medulla, and development of the cortical ridge (Chang et al. 2019). These factors can vary along the proximal-distal axis of a single feather, among feathers from different body regions of a single taxon, and among taxa to help birds adapt to different ecological niches (Chang et al. 2019).

However, an unexpected feather morphotype reported in several lineages of nonavian and avian theropods (Zhang et al. 2008a; Zhang et al. 2008b; Ji et al. 1999; Zhang and Zhou 2000; O’Connor et al. 2012; Wang et al. 2014), known as rachis-dominated feathers (RDFs), is characterized by an unusually wide and flattened rachis with a pair of lateral light stripes delimiting a darker medial stripe (O’Connor et al. 2012; Wang et al. 2014). The rachis of RDFs can be bordered laterally by pennaceous vanes that either extend throughout the length of the feather (e.g., Eopengornis [Wang et al. 2014]), are restricted to the distal end (e.g., Confuciusornis [Hou et al. 1995]), or are devoid of vanes (e.g., Epidexipteryx [Zhang et al. 2008b]). Accordingly, RDFs have been referred to as rachis-dominated pennaceous feathers (O’Connor et al. 2012; Wang et al. 2014), rachis-dominated “racket-plumes” (sensu proximally ribbon-like pennaceous feather) (Foth 2012; O’Connor et al. 2012; Wang et al. 2014), and ribbon-like feathers (Zhang et al. 2008b), respectively. The detailed structures of these distinct rachises have traditionally been obscured by the carbonized nature of the compressed fossils (O’Connor et al. 2012). Fortunately, recent studies of isolated feathers embedded in early Late Cretaceous Burmese ambers (about 98.8 ± 0.6 million years ago, Cenomanian [Shi et al. 2012]) have revealed that the rachises of RDFs are not cylindrical (Xing et al. 2018; Carroll et al. 2019). Instead, these rachises consist of only the dorsal cortex, lacking the medulla and ventral cortex, and therefore the rachises appear as if they are ventrally opened (Xing et al. 2018; Carroll et al. 2019). This unique structure raises the questions of how a rachis without a medulla and a ventral cortex could be formed, and whether this configuration represents a distinct phenotype that evolved independently of the common cylindrical rachidial phenotype (Carroll et al. 2019).

Here, we worked to answer these questions using approaches that combine advantages of paleontology, functional simulations, and developmental biology. Paleontologically, a total of seven isolated RDFs preserved in early Late Cretaceous Burmese ambers are described and categorized into three distinct morphotypes based on the morphologies of the ventrally open rachises. The relative stiffness of feathers with various rachidial morphologies was then estimated through functional simulations to test the effects of ventrally open rachises on feather strength, toward clarifying feather functions, and adaptive limits and advantages. Because feathers undergo cyclic regeneration, whereby the distal portion of feather matures first and the proximal portion is still undergoing morphogenesis, this allows a temporal developmental process to be laid out along the proximal (immature)-distal (mature) axis of a regenerating feather, which facilitates the study of the morphogenetic process. To understand the developmental origin of each morphotype of RDFs, we examined cell proliferation and keratin differentiation in the regenerating rachises of the extant chicken (Gallus gallus). By examining the feathers of extant penguins, we suggest that a ventrally open rachis also characterizes the remiges and rectrices of adult Pygoscelis papua, which indicates that the RDFs may not be unique to the Mesozoic feathers.

Materials and Methods

ANCIENT AND EXTANT FEATHERS

All amber specimens described in this article (Figs. 1 and S1) are housed in the College of Life Sciences, Capital Normal University, Beijing, China (CNU). A regenerating breast contour feather and a 4-week-regenerating fifth primary remex used for histological sections were extracted from a one-year old White Leghorn chicken (G. gallus). Feathers of extant penguins were taken from taxidermy of a mature Pygoscelis papua (CNU-Z1025) housed in the Zoology Collection, Capital Normal University, China, and an uncatalogued taxidermy housed in the National Museum of Matine Biology and Aquarium, Taiwan. Extraction and sectioning were performed followed by the approved protocol of Capital Normal University, China, and China Medical University, Taiwan, respectively.

Figure 1.

Figure 1.

Open in a new tab

Unusual rachidial morphotypes of isolated RDFs preserved in the early Late Cretaceous Burmese ambers. Left, schematic drawings showing the general morphologies (left) and the rachidial cross-sections (right) of morphotype IA (A), IB (B), and II (C) RDFs, respectively (not to scale); white arrowheads mark the cortical ridge. Right, photos showing the amber embedded RDFs and rachidial cross-sections: (D) photo of CNU A0002 showing the rachis of morphotype IA RDFs; red dashed lines and box mark the positions of the cross-sections and the close-up images shown in (E-H), respectively; (E-G) cross sections of CNU A0002 at different positions; white arrowheads mark the cortical ridge; (H) close-up image showing the open rachis of CNU A0002; (I-L) photos of CNU A0008 (I, J) and A0009 (K, L) showing rachises of morphotype IB RDFs. Red boxes in (I) and (K) mark the close-up images shown in (J) and (L), respectively; (M) photo of CNU A0005 showing the rachis of a morphotype II RDFs; the red box and the dashed line mark the positions of the cross-sections and the close-up image in (N) and (O), respectively; (N) close-up image of CNU A0005 showing rachidial light and dark stripes; (O) cross-section of CNU A0005 showing the boomerang-shaped dorsal cortex.

Abbreviations: af = afterfeather; ca = calamus; de = distal expansion; lf = lateral flange; ra = rachis.

Scale bars for (D, I, K, and M) = 2 mm; scale bars for the rest panels = 0.2 mm.

HISTOLOGICAL SECTIONS AND STAINING

For ground sections, each amber was cut perpendicular to the long axis of the rachis to an appropriate thickness (usually 100–120 μm), using an EXAKT 300CP automatic microtome with a 0.35-mm-wide blade (EXAKT-300CL, EXAKT Band System). The cut sections were then polished using an EXAKT 400CP variable speed grinder-polisher successively with P800 and P1200 polishing paper until their thickness was reduced to a final value of 40–60 μm, facilitating examination by transmitted and light microscopy (Leica DMRX). Detailed images of histological slides were obtained with a Nikon DS-Ri 2 Camera System through a Nikon SMZ25 transmitted light microscope.

Rachises of the regenerating chicken contour and primary remex were embedded in paraffin and sectioned every 2 and 10 mm, respectively, from the immature proximal end to establish growth series. The samples were fixed in 4% paraformaldehyde for 24 h, and then dehydrated using a graded ethanol series from 30% to 100%. Paraffin sections (10-μm thick) were cut perpendicular to the long axis of the rachis. H&E staining, BrdU assays, and immunohistological staining were performed according to standard protocols described by Schulte (1991), Lee et al. (2001), and Chang et al. (2019), respectively. Polyclonal antibody against feather keratin encoded on Chromosome 27 of G. gallus (Wu et al. 2015) was generated by Gen-Script, using the SEGVPITSGGFDLSC epitope sequence.

FUNCTIONAL SIMULATIONS

Tensile strengths of feathers with eight different combinations of rachidial and vane morphologies in response to a standardized external simulated force (50 N) was estimated using Solid-Works (Fig. S2), and the displacement and the static stress of each model were analyzed. The hollow cylindrical rachises simulate medulla-free feathers with fully developed (e.g., extant penguin [Chang et al. 2019]; Figs. 2A and S3A) and distally located vanes (Figs. 2B and S3B); the solid cylindrical rachis with fully developed vanes simulates the ideal remiges of extant birds (Figs. 2C and S3C); the solid cylinder rachis with distally located vanes simulates the ribbon-like rectrices in extant Prioniturus discurus (Figs. 2D and S3D) (Bleiweiss 1987); the crescent-shaped rachis with a rachidial ridge and fully developed vanes simulates the rectrices of Eopengornis (morphotype II RDFs; see below for a detailed interpretation) (Figs. 2E and S3E); the crescent-shaped rachis with a rachidial ridge and distally located vanes simulates the rectrices of Confuciusornis (morphotype II RDFs; see below for a detailed interpretation) (Figs. 2F and S3F); and the proximally crescent-shaped rachis with and without a rachidial ridge simulates morphotype IA (Figs. 2G and 3G) and IB (Figs. 2H and S3H) RDFs, respectively. In these models, the proximal end of the rachis is fixed to simulate attachment of the feather in the skin, so only the distal end is free. We assume that all forces are evenly applied to vanes and then pass to the rachis, and all barbs attached to the rachis are interlocked to form a complete vane.

Figure 2.

Figure 2.

Open in a new tab

Simulations showing displacement of feathers with various combinations of vane and rachidial morphologies. Resultant displacement (URES) is expressed in mm. Models simulate a hollow rachis with fully developed vanes (A); a hollow rachis with distally located vanes (B); a solid rachis with fully developed vanes (C); a solid rachis with distally located vanes (D); morphotype II RDFs with a rachidial cortical ridge and fully developed vanes (E); morphotype II RDFs with a rachidial cortical ridge and distally located vanes (F); morphotype IA RDFs with a rachidial cortical ridge and fully developed vanes (G); and morphotype IB RDFs with fully developed vanes but without a rachidial cortical ridge (H). The symbol underneath each model represents the type of rachidial cross section. The results show that models simulating cylindrical rachises filled with medulla (C and D) are more stable than other rachidial architectures.

Figure 3.

Figure 3.

Open in a new tab

Tissue differentiation (H&E), cell proliferation (BrdU), and keratinization (polyclonal antibody) of a 4-week regenerating remex from a one-year old chicken. (A) Photo of the regenerating remex rachis. Red box shows an enlargement of the proximal feather region, and the yellow lines mark positions of the slices shown in (B-G); (B-G) H&E, (B, C) BrdU, (D, E) and feather keratin immunohistochemistry (F, G) staining showing the rachidial morphology, cell proliferation, and differentiation at each level of a regenerating remex rachis. Yellow boxes mark the enlarged areas of the corresponding images. The white dashed line marks the epidermal-pulp boundary. Scale bars = 2 mm for (A); 0.5 mm for (B, D, F); 0.2 mm for (C, G); and = 0.1 mm for (E).

Simulations of each morphotype were subject to a standardized force of 50 N. This standardized force (50 N) is exaggerated beyond the maximum force a normal extant contour feather can withstand, to reveal where the broken position of the rachis is likely to occur (Fig. S3, red arrows). Although the absolute values in each test may not reflect real conditions, the relative flexing (displacement) and the static stress in response to the standardized external simulated force are comparable to each other, reflecting the relative stiffness of the different feather morphotypes.

The parameters of the analyzed material were set to the physical properties of real feather keratins, with the material density of 1.15 g/cm3 (Hertel 1966), and Young’s modulus (a parameter characterizing the material stiffness against the external force) of 2.5 × 109 N/M2 (Macleod 1980; Bonser and Purslow 1995). The model feather is 200-mm long in all simulations, and only a 3.5-mm-wide vane was set on each side of the rachis because all force must pass from the proximal end of the barbs to the rachis (Figs. S2C, S2D, S2G, and S2H). For models A-D, the cylindrical rachis is 2.5 mm in diameter and the cortex thickness is 0.3 mm if the rachis is not solid (Figs. S2A-B). Models simulating rachis with well-developed vanes have a 20-mm calamus and therefore the remaining 180-mm rachis attaches to the vanes (Fig. S2E), whereas models simulating rachis with distally located vanes only attach to a pair of 40-mm long vanes distally (Fig. S2F). The rachis distal to the ventrally open region is represented by Φ = 0.15 mm (Φ, diameter) solid beam in models simulating the morphotype IA and IB RDFs (Figs. S2G and S2H). The ventrally open rachis corresponds to one-third of the cortex of the hollow rachis, which extends 90 mm (Figs. S2C, S2D, S2G, and S2H).

Results

MORPHOTYPES OF THE RDFs

The RDFs present here are divided into three distinct morphotypes based on the extent of the ventrally open rachis and the presence or absence of the midline cortical ridge.

Morphotype I RDFs (Figs. 1A and 1B) are represented by five specimens: CNU A0002 (Fig. 1D-H), A0007 (Figs. S1A and S1B), A0008 (Figs. 1I and 1J), A0009 (Figs. 1K and 1L), and A0010 (Figs. S1C and S1D). Each specimen contains an incomplete isolated feather with symmetrically distributed barbs and long barbules. Unlike the common tubular feather rachises, the proximal half of each feather rachis is transversely expanded and appears fusiform in shape when viewed anteriorly (Figs. 1A, 1B, 1D, 1H, 1I, and 1K). Cross sections show that the proximal rachis consists of only a crescent-shaped dorsal cortex with two lateral flanges, and that each flange ends distally with a drop-like expansion (Fig. 1F-G). Besides the dorsal cortex, there is no sign of medulla or ventral cortex, and therefore the proximal part of the rachises is ventrally opened (Figs. 1F, 1G, and 1O).

The width of the ventrally open region can vary along the feather (Figs. 1A and 1B), and among specimens, reflecting varied development of the dorsal cortex. Where the ventrally open rachis occurs, the dorsal cortex is fairly thin, and the barbs fuse into the rachis at the base of the distal expansions (Fig. 1E-G). Therefore, morphotype I RDFs can be diagnosed by a proximal ventrally open rachis that tapers to a solid beam before reaching the distal tip (Fig. 1E; Table S1). In two specimens (CNU A0002 and A0007), the lateral flanges separate from each other by a midline cortical ridge that protrudes ventrally from the dorsal cortex (Figs. 1D and 1H). Although the cortical ridge extends throughout the ventrally open region in CNU A0007, this ridge is restricted to the proximal three fourths of the ventrally open region in CNU A0002, and completely absent in the other three specimens (CNU A0008, A0009, and A0010; Figs. 1J, S1C, and S1D), producing three feather morphotypes that are unknown in extant feathers. A dorsal groove corresponding to the midline ridge reported in a previous work (Carroll et al. 2019) is absent in any of the morphotype I specimens presented in this article. To understand the developmental etiology of the absent medulla and ventral cortex, we further subdivided morphotype I RDFs into two subcategories based on whether a midline cortical ridge is present (morphotype IA; Fig. 1A) or not (morphotype IB; Fig. 1B). The previously reported Burmese amber-embedded RDFs DIP-V-15125 and DIP-V-16186 (Xing et al. 2018) are tentatively assigned by us to morphotype IB RDFs (Table S1).

Morphotype II RDFs (Fig. 1C) are represented by two specimens: CNU A0005 (Fig. 1M-O) and A0011 (Fig. S1E-H). CNU A0005 contains a nearly symmetric isolated feather with barbs and barbules that are shorter and denser than those of the morphotype I RDFs (Fig. 1M). The preserved portion indicates this feather could be no less than 8-cm long, but a considerable proximal portion of the feather is missing. Unlike morphotypes IA and IB, the rachises of morphotype II RDFs have ventrally open regions that extend all the way to the distal end of the feather where the barbs and barbules are too dense to discern detailed structures (Figs. 1M and 1N). The lateral flanges are extremely thin and arch dorsally, giving the rachis a boomerang-shaped appearance in cross section view (Fig. 1O). From the cortex protrudes a midline ridge that is approximately four times as thick as the lateral flange (Fig. 1O), but again a dorsal groove corresponding to the midline ridge is unknown in either of the morphotype II specimens presented here. Therefore, in anteroposterior view the rachis is ribbon like rather than fusiform as in the morphotype I RDFs. This phenotype resembles the ribbon-like feather rachises documented in several lineages of avian and nonavian theropods (Zhang et al. 2008b; O’Connor et al. 2012; Wang et al. 2014) and indicates that the light and dark stripes that characterize the rachises of RDFs in the compressed fossils could correspond to the thin lateral flanges and cortical ridge of the ventrally open rachises.

The rachis of CNU A0011 is much thinner than that of CNU A0005, but histological sections reveal similar rachidial configuration except for an unusually thick cortical ridge that is either mushroom like or bell shaped in cross section depending on the location of the section (Fig. S1F-H). Combining these features, morphotype II RDFs can be diagnosed by a ventrally open rachidial region accompanied by the cortical ridge that extends all the way to the distal end of the feather (Fig. 1C). Most previously reported Burmese amber-embedded RDFs (Xing et al. 2018) and all known RDFs recognized in the compressed fossils belong to this morphotype II (Zhang et al. 2008b; Wang et al. 2014) (Table S1).

COMPARISONS OF RACHIS STRENGTH

Variability in feather rachis flexural stiffness has been suggested to be determined by cross-sectional geometry, rather than via physical properties of feather keratins (Bonser and Purslow 1995; Wang and Meyers 2017; Chang et al. 2019). Stiffness, in this context, is the resistance to deformation in bending and/or torsion. The strength of a beam refers to the force needed to break the beam (in this case, a rachis) by bending or twisting. The simulation results show that the maximum flexing of RDFs with a given length could be up to 100 times less resistant to flexing than feathers with cylindrical rachises (Fig. 2), suggesting that the ventrally open rachises are mechanically unstable and easier to twist and bend than the cylindrical rachises. Although some flexibility along the rachis can make a feather less likely to fracture (by absorbing energy or impacts), such extreme flexing would likely render a feather aerodynamically unstable and severely compromise performance under high loads. Low stiffness in bending, relative to twisting, is particularly likely to reduce feather performance. Models simulating morphotype I RDFs are most easily twisted and fractured regardless of whether a cortical ridge is present or not (Fig. 2G, H). In contrast, model simulating the typical cylindrical rachises filled with medulla is the most stable configuration (Fig. 2C). These results suggest the variety of rachidial configurations assessed, for a given length and diameter, differs substantially in stiffness with the cylindrical medulla-filled rachises being most resilient to stress and strain.

DEVELOPMENT OF EXTANT FEATHER RACHISES

The H&E and BrdU staining of chicken feather serial sections demonstrated that the initial feather rachis development is characterized by the formation of dorsal cortex (Stage I), followed by the proliferation of basal epidermis cells that are tightly appressed to the inner surface of dorsal cortex (Fig. 3E, Stage I). Cell proliferation does not occur in specific clusters, nor does it occur evenly across the basal epidermis, because our analysis shows that groups of BrdU positive cells are separated intermittently by low-proliferating zones (Fig. 3D-E, Stage I). Cell differentiation occurs soon after they proliferate, as cells generated from the high-proliferation zones are densely grouped and became fusiform with their long axes orienting toward the center of the follicle, forming cortical ridge precursors (Fig. 3, Stage II). This makes the high-proliferation zones grow faster toward the center of the follicle than the low-proliferation zones, giving the ventral margin of the developing rachis a wave-like appearance in cross section (Fig. 3, Stages I and II). In Stage II and more mature rachidial sections, cell proliferation diminishes as cell differentiation becomes predominant. Cells derived from the low-proliferation zones become enlarged and form the initial spongy medulla (Fig. 3, Stage II). In the following stages, these differentiating cells express beta-keratin transiently. Then they become medulla cells with specific size and arrangement, forming the major bulk of the medulla (Fig. 3B-D). Meanwhile, the dorsal cortex extends all the way around the developing medulla, and the lateral and the ventral rachis cortexes form successively (Fig. 3B). Therefore, the feather rachis develops ventrally from the dorsal cortex.

The keratinization process of the rachis is evidenced by feather keratin immunohistochemical staining (Figs. 3D-G), whereby differentiated cells die soon after maturation and become keratinized. Unsurprisingly, feather keratin is first expressed in the dorsal cortex (Figs. 3D-E), and cortical ridge keratinization is completed when medulla differentiation initiates (Fig. 3C). The spongy medullary cells then become keratinized vacuoles that strengthen the cylindrical external cortex (Fig. 3B-C). Finally, the entire rachis is completely keratinized when the keratinization of the ventral cortex is achieved. Therefore, the keratinization of the feather rachis follows a similar spatiotemporal order to cell proliferation. Morphogenic processes of regenerating covert contours are basically the same as those of the remex (see Supporting Information for detailed description), but they are not as clear as those in the remex due to the smaller sizes of contour feathers (Fig. S4). Regulatory molecules involved in these differentiation processes were reported elsewhere (Chang et al. 2019).

Discussion

TAPHONOMY EFFECTS

The origin and evolution of feathers are becoming more clear as exquisitely preserved avian- and nonavian, feathered dinosaurs are continuously uncovered from Mesozoic deposits (McKellar et al. 2011; Xing et al. 2018; O’Connor 2020; Xu 2020). Despite their exceptional preservation, the integuments preserved in compressed fossils are usually taphonomically altered such that detailed feather morphologies are obscured or unpreserved (Xing et al. 2017; Yang et al. 2019). Thus, amber-embedded RDFs provide rare opportunities to investigate detailed morphologies of early feathers.

Feather keratin is extremely resistant to water and organic solvents (Moyer et al. 2016; McCoy et al. 2019). Although fossilization in resin flows can disturb the original orientation of constituent feather branches such as barbs and barbules, the rachidial structures remain mechanically unaltered and are usually better preserved than other parts of the feather (Zhao et al. 2020). Even if buried amber-embedded feathers had been subjected to mechanical pressures as strong as those of compressed fossils after they were buried, there is no reason to believe the rachidial dorsal cortex, barbs, and even barbules are more durable than the ventral cortex and medulla. In addition, the medulla is also absent in the distal solid beam of morphotype I RDFs (Fig. 1E), implying that taphonomic factors are unlikely to be responsible for the absent rachidial medulla and ventral cortex in the amber embedded RDFs.

TAXONOMIC AFFINITY AND THE DISTRIBUTION OF RDFs

A rachis-like structure is thought to have evolved as early as in Triassic Avemetatarsalian ancestors (Yang et al. 2019; Xu 2020), and a true rachis was first seen in maniraptorans (Xu and Guo 2009). Among theropods, feathers with ventrally open rachises are primarily documented in remiges and rectrices of many paravians (e.g., RDFs present in Confuciusornis [Hou et al. 1995] and Eopengornis [Wang et al. 2014]; Fig. 4). However, it is hard to determine whether RDFs are also present in covert contours because contour feathers are usually too dense to be discerned in most compressed specimens, and many ventrally open rachises are too thin to be identified (e.g., CNU A0011). Although several exquisitely 3-dimensional preserved feathered theropods were reported from the Burmese ambers recently (Xing et al. 2016a; Xing et al. 2016b; Xing et al. 2017), little is known about the ventrally open rachis in any of the feathers attached to these theropods, and no high-resolution computerized tomography (CT) images are currently available to verify these observations.

Figure 4.

Figure 4.

Open in a new tab

Distribution of rachis-dominated feathers (RDFs) across a simplified cladogram of theropod modified from (Xu and Guo 2009; Jarvis et al. 2015; Wang et al. 2015). Lineages without marking feather symbols indicate feather morphologies are unknown.

Because the inclusion size is severely limited by the generally small resin droplet size, isolated feathers are more commonly preserved in ambers than those attached to the animal remains (Perrichot et al. 2008; McKellar et al. 2011; Xing et al. 2020). This results in the uncertainty regarding their taxonomic affinities. In extant birds, contour feathers cover most of the body surface and serve to streamline it for flight and prevent heat loss, among other functions (Lucas and Stettenheim 1972; Stettenheim 2000). Remiges and rectrices are more strongly modified for flight and differ from covert contour feathers in having relatively larger sizes and aspect ratios, stiffer rachises, hooklets on the distal barbule end, and asymmetric vanes (especially for remiges) that are almost entirely pennaceous (Lucas and Stettenheim 1972). Assuming differentiation of Mesozoic feathers is generally identical to that of extant ones, the above amber-embedded RDFs except CNU A0005 are likely to be covert contours judging from their general morphologies. On the other hand, CNU A0005 resembles more closely the less developed symmetric flight feathers given its large aspect ratio and an entirely ventrally open rachis, which is documented more often in remiges and rectrices in the fossil record. However, interpreting more specific body locations as the source for the isolated amber-embedded RDFs of uncertain taxonomic affinities is impossible, as contours with various rachidial morphotypes may come from different feather tracts, ages, or sexes. Similarly, we choose not to interpret the growth status of each isolated amber-embedded RDFs because the growth rate and duration of each feather may be different even in extant birds, depending on the exact position and function of the particular feather.

Besides Mesozoic feathers, the RDFs with ventrally open feather rachises are also known in a few extant birds (Rutschke 1965; Lucas and Stettenheim 1972). In penguins (e.g., Aptenodytes forsteri [Rutschke 1965]; Pygoscelis papua [Chang et al. 2019]), although some covert contour feather rachises are enriched with medulla, the medulla and ventral cortex are absent from the proximal half of the rectrices and remiges in a way that resembles those of morphotype I RDFs (Fig. S5; see Supporting Information for more discussion) (Chang et al. 2019; Foth 2020). The absence of medulla in the modified flight feathers reduces buoyancy while a penguin dives, whereas the covert contour feathers with a large bulk of medulla prevent the loss of heat. Because such regional-specific feather morphological differences on the same bird can be regulated epigenetically in extant birds (Jiang et al. 2004; Chen et al. 2015), we speculate that a similar regional specificity could also account for the presence or absence of RDFs in the fossil records.

VARIATIONS IN FEATHER STIFFNESS

Flexural stiffness and curvature are two key factors affecting feather stability and functions (Bachmann et al. 2012; Wang and Meyers 2017). Flight feathers must be strong enough to withstand large aerodynamic forces. On the other hand, although covert contour feathers do not need to withstand aerodynamic forces of such large magnitude, they need to maintain enough mechanical strength to keep the body surface streamlined during flight and/or to prevent heat loss (Stettenheim 2000; Gill et al. 2019). Even when serving communication purposes (i.e., attraction or intimidation), covert contour feathers on the neck of chickens are often held erect (Gill et al. 2019), and hence a certain level of mechanical strength is still necessary.

Some estimation can be made regarding the force a flight feather experiences during flight. The lifting force for a bird during flight is mainly contributed by about 40 plus flight feathers (primary and secondary feathers on the left and right wings). To counter gravity during sustained nonflapping flight, the minimum load acting on a flight feather should, therefore, be at least 1/40 of the body weight. For a 1-kg chicken, the force for each single flight feather should be at least 0.25 N during its flight. For a finch weighing about 0.02 kg, each feather would experience about 0.005 N, whereas each feather of a bald eagle weighing about 5 kg would experience 1.25 N. Feathered nonavian dinosaurs and most stem birds tend to be larger and heavier, and hence the force their flight feathers must experience would be even larger (Benson et al. 2014; Benson et al. 2018). To discuss the matter in further detail, the amount of force exerted on a specific point of the rachis depends on its exact position in the wing plane, and the magnitude of this force also fluctuates dynamically during wing flapping. These rough estimates represent minima, as aerodynamic loads on the feathers are much greater during flapping flight, especially during a steep climb.

The bending behavior of the feather rachis is largely determined by the stiffness of the keratin composite and the shape of the rachidial cross sections (Bonser 1996). Feather components, including cortex, medulla, barbs, and barbules, are primarily made of beta-keratin (Schor and Krimm 1961); previous studies have found little interspecific variation in Young’s modulus of the cortex (Bonser and Purslow 1995), although that of the medulla is relatively low (Bonser 1996). Our functional simulations corroborate the idea that for a given diameter, a more solid rachis is more resistant to torsion and bending (Figs. 2 and S3), and the flexural stiffness of a beam is highly dependent on its diameter. In other words, larger diameter beams with thin walls are stiffer (and stronger) than smaller diameter beams composed of the same material, even if the smaller diameter beams are solid (see Supporting Information).

However, real feathers (in both living and fossil taxa) vary in rachidial configurations (Bachmann et al. 2012; Wang and Meyers 2017; Chang et al. 2019). A ventrally open rachis has a smaller effective diameter (lower polar moment) than a complete tube. In addition, a ventrally open rachis can further open under bending deformation regardless of whether a differentiated medulla is present or not, and in this condition the rachis would begin to act more like a curved plate than a tube (e.g., fig. 2D in Xing et al. 2018). The effective diameter of a curved plate is closer to its thickness (which is small) than to the outer dimension of the curvature (Bachmann et al. 2012). Therefore, the stiffness of a ventrally open rachis can be increased only through widening of the rachidial diameter, resulting in the fusiform appearance of the ventrally open region.

Furthermore, resistance to torsion is proportional to the polar moment of inertia, which effectively means that twisting stiffness is proportional to the “average” diameter. The large groove in the ventrally open rachis produces an area with low effective diameter, greatly decreasing the average diameter and the resistance to twisting. In cylindrical feathers, ventral grooves are used to provide a high ratio of twisting to bending. The ventrally open rachis would have had a high ratio of twisting to bending, as well, but the morphology is overall much weaker than the cylindrical feathers, especially in bending, because of the tendency of the feather rachis to open and fail rapidly when plastically deformed.

Cylindrical feathers of extant birds obtain high flexural stiffness not by being truly solid, but by having a large cortex diameter supported by an internal medullary foam (Lucas and Stettenheim 1972; Chang et al. 2019). Expanding the cortical diameter in the feather rachis provides increased flexural stiffness and strength (Bachmann et al. 2012; Wang and Meyers 2017). By filling the center of the rachis with a foam-like medulla, increased flexural stiffness comes at a very low cost in added featherweight. The simulations successfully predict that the rachises with only the dorsal cortex are less stiff than those that form a complete cortex filled with medulla, which indicates the cylindrical rachises would likely have first evolved to buffer against the mechanical weakness of the RDFs.

PUTATIVE DEVELOPMENTAL MECHANISM OF THE VENTRALLY OPEN RACHIS

Our results show that the rachidial dorsal cortex, cortical ridges, medulla, and ventral cortex form sequentially in extant feathers. From a developmental perspective, the absence of rachidial medulla and ventral cortex in RDFs could be achieved through either an early truncation of cell proliferation (which likely cause a complete loss of medulla and ventral cortex) or absence of keratinization of differentiated medullary tissue (both medulla and ventral cortex are differentiated but lack keratinization) or the combination of both.

The various extents of the ventrally open region and the cortical ridge in the specimens deserve a special note. Assuming that the developmental trajectories of early feathers are identical to those of extant ones, the presence of the cortical ridge indicates that the medulla should have differentiated to some extent (Fig. 5). From this perspective, the open region of morphotype IA RDFs, where a cortical ridge is present, is developmentally no younger than the Stage II developing rachis of extant feathers (Fig. 5). The rachidial development in morphotype IA RDFs (e.g., CNU A0002) feathers thus can be depicted in the following scenarios: (i) when a feather starts growing, dorsal cortex differentiation is initiated, forming the distal solid beam lacking dorsal cortex expansion and differentiations of all other rachidial tissues (Fig. 5; this stage could be even younger than the Stage I developing cylindric rachis of extant feathers); (ii) the dorsal cortex expands in width but lacks further cortical ridge differentiation, producing the middle rachis that is equivalent to the Stage I developing cylindric rachis of extant feathers (Fig. 5); (iii) the rachidial tissue further differentiates when the proximal half of the feather starts growing. Although the dorsal cortex and cortical ridge become fully keratinized, a failure to keratinize of the medulla and the ventral cortex would result in the proximal half of the open rachis being morphologically equivalent to the Stage II developing cylindrical rachis (Fig. 5). If the medulla and ventral cortex completely failed to keratinize, it could form the rachises in morphotype IB RDFs that completely lack the cortical ridge (Fig. 5).

Figure 5.

Figure 5.

Open in a new tab

Schematic drawings showing that diverse rachidial morphologies of RDFs are equivalent to various developmental stages of extant feather rachis. Not to scale.

For the rachises of morphotype II RDFs, tissue differentiation is more advanced than morphotype I RDFs and is relatively homogeneous throughout feather growth, resulting in the entire rachis that is morphologically equivalent to the Stage III developing cylindrical rachis (Fig. 5). Therefore, the rachis of morphotype II RDFs is developmentally more mature than that of morphotype I RDFs and is likely to have been mechanically stronger to withstand the airflow as they are frequently present on remiges and rectrices (O’Connor et al. 2012; Wang et al. 2014). The width expansion of the dorsal cortex and the formation of medulla have recently been verified to be regulated by BMP4 and Ski (a TGFβ inhibitor), respectively (Chang et al. 2019), but the mechanisms for the absence of the posterior rachidial tissue remain unclear. Taken together, current lines of evidence support the idea that a ventrally open rachis is developmentally equivalent to the immature cylindrical rachis (Fig. 5).

THE PREVALENCE OF THE CYLINDRICAL RACHISES

The various developmental degrees of rachidial tissue produce a series of continuous characters that are likely to result in discriminate fitness (e.g., the loss of the rachidial medulla in extant penguins is more effective in reducing the buoyancy, whereas the increase of the rachidial width is likely to further strengthen the feather in case where the medulla is lost).

The feather morphotypes present in crown birds evolved from their Cretaceous ancestors. But the knowledge gap on feather morphology of the most recent common ancestor of crown birds has significantly impeded our understanding on the rachidial complexity in extant birds. Although a variety of rachidial configurations have been documented in Mesozoic theropods, the global catastrophe that occurred at the end of the Cretaceous (∼66 million years ago) wiped out all nonavian and most avian theropods (Brusatte et al. 2015a). Some feather morphotypes present in the Cretaceous archaic birds and nonavian theropods hence have also gone extinct. This led to only a few lineages of avian theropods that survived past the Cretaceous and into the Paleogene (Brusatte et al. 2015b), leaving them as the only representatives of the initial feather diversification.

The ecological release afforded by the extinction of their Mesozoic relatives is thought to have allowed the Paleogene bird survivors to diversify rapidly (Brusatte et al. 2015b; Field et al. 2018). However, the cylindrical configurations of the feather rachis have dominated the crown birds including groups that have secondarily lost flight capabilities (e.g., kiwi and ostrich), regardless of the ecological niches they have occupied. This leads to a parsimonious assumption that all bird lineages that survived the extinction event have feathers with cylindrical rachises, although it is unclear why birds with cylindrical rachises are more likely to have survived. Alternatively, only paleognaths inherited the cylindrical feather rachises from the Cretaceous ancestor, whereas the evolution of the cylindrical rachises in other clades of crown birds was driven independently by flight requirements. However, the latter explanation is less likely given the rapid adaptive radiation of birds in the postapocalyptic world, and no example of the ventrally open rachis has been documented in the Cenozoic fossil records (Mayr 2009; Nesbitt and Clarke 2016). Therefore, current evidence implies that the prevalence of the cylindrical configurations of feather rachises in most crown birds is likely to be maintained by developmental constraints inherited from the Cretaceous ancestors. This further corroborates the idea that the spatiotemporal-specific presence of the ventrally open rachis could have evolved independently in extant penguins to adapt the semiaquatic lifestyle. Therefore, the ventrally open rachises present in extant penguins may not be achieved through the same mechanism as the amber-embedded RDFs (see Supporting Information for more discussion), although they morphologically resemble each other. The morphogenesis of RDFs in extant penguins remains an open question for future studies.

Conclusions

Our work suggests that the rachises of fossil RDFs are equivalent to the poorly developed extant cylindrical rachises. Based on the extent of the ventrally open rachis and the presence or absence of the midline cortical ridge, the known fossil RDFs are divided into three distinct morphotypes, whereby morphotype II may be more mature than morphotypes IA and IB, which may correspond to the Stage III developing extant cylindrical rachis. Mechanical strength is likely to have been a key factor that drove the evolution of cylindrical rachises, and the prevalence of cylindrical feather rachises in crown birds is likely to be maintained by developmental constraints inherited from the Cretaceous ancestors.

Supplementary Material

Electronic supplementary information text, Figs. S1 to S5, Table S1, and References for the electronic supplementary information.

Table S1. Summary of RDFs preserved in Burmese ambers.

Fig. S1. Photos of RDFs preserved in the early Late Cretaceous Burmese ambers.

Fig. S2. Schematic drawings showing the parameters using in the functional simulations.

Fig. S3. Static stress analyses of feathers with various combinations of vane and rachidial morphologies.

Fig. S4. Tissue differentiation (H&E) of a regenerating contour feather from a one-year old chicken.

Fig. S5. RDFs in extant penguins (Pygoscelis papua).

Supporting Information

Additional supporting information may be found online in the Supporting Information section at the end of the article.

ACKNOWLEDGMENTS

We wish to thank the editors and two anonymous reviews for their constrictive critiques and advices improving the manuscript. SW is supported by the Human Frontier Sciences Project (LT000728/2018); PW and C-MC are supported by the National Institutes of Health, USA (AR47364, AR60306) and a research contract between China Medical University (CMU)/ Hospital in Taiwan and USC (USC grant number 5351285884). C-MC is a paid scientific advisor of CMU/ hospital. W-LC and W-TJ are supported by the Integrative Stem Cell Center, CMU/ Hospital in Taiwan, grants from Taiwan Higher Education Sprout Project by the Ministry of Education in Taiwan, and grants from Ministry of Science and Technology in Taiwan. We thank D. Ren and T. Gao (CNU) for proving important samples and helping photography, A. Li, P. Zhang, H. Zhang (Capital Normal University) and H. Mao (Peking University) for discussing the functional simulation results, and R. Widelitz for improving the manuscript.

Footnotes

CONFLICT OF INTEREST

The authors declare no conflict of interest.

DATA ARCHIVING

Data are available on the Dryad Digital Repository (https://doi.org/10.5061/dryad.18931zctm).

LITERATURE CITED

  1. Bachmann T, Emmerlich J, Baumgartner W, Schneider JM, and Wagner H. 2012. Flexural stiffness of feather shafts: geometry rules over material properties. J. Exp. Biol 215:405–415. [DOI] [PubMed] [Google Scholar]
  2. Benson RBJ, Campione NE, Carrano MT, Mannion PD, Sullivan C, Upchurch P, and Evans DC. 2014. Rates of dinosaur body mass evolution indicate 170 million years of sustained ecological innovation on the avian stem lineage. PLoS Biol. 12:e1001853. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Benson RBJ, Hunt G, Carrano MT, Campione N, and Mannion P. 2018. Cope’s rule and the adaptive landscape of dinosaur body size evolution. Palaeontology 61:13–48. [Google Scholar]
  4. Bleiweiss R 1987. Development and evolution of avian racket plumes: fine structure and serial homology of the wire. J. Morphol 194:23–39. [DOI] [PubMed] [Google Scholar]
  5. Bonser R, and Purslow P. 1995. The Young’s modulus of feather keratin. J. Exp. Biol 198:1029–1033. [DOI] [PubMed] [Google Scholar]
  6. Bonser RHC 1996. The mechanical properties of feather keratin. J. Zool 239:477–484. [Google Scholar]
  7. Brusatte SL, Butler RJ, Barrett PM, Carrano MT, Evans DC, Lloyd GT, Mannion PD, Norell MA, Peppe DJ, Upchurch P, et al. 2015a. The extinction of the dinosaurs. Biol. Rev 90:628–642. [DOI] [PubMed] [Google Scholar]
  8. Brusatte SL, O’Connor JK, and Jarvis ED. 2015b. The origin and diversification of birds. Curr. Biol 25:R888–R898. [DOI] [PubMed] [Google Scholar]
  9. Carroll NR, Chiappe LM, and Bottjer DJ. 2019. Mid-Cretaceous amber inclusions reveal morphogenesis of extinct rachis-dominated feathers. Sci. Rep 9:18108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Chang W-L, Wu H, Chiu Y-K, Wang S, Jiang T-X, Luo Z-L, Lin Y-C, Li A, Hsu J-T, Huang H-L, et al. 2019. The making of a flight feather: bio-architectural principles and adaptation. Cell 179:1409–1423. e1417. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Chen C-F, Foley J, Tang P-C, Li A, Jiang TX, Wu P, Widelitz RB, and Chuong CM. 2015. Development, regeneration, and evolution of feathers. Ann. Rev. Anim. Biosci 3:169–195. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Field DJ, Bercovici A, Berv JS, Dunn R, Fastovsky DE, Lyson TR, Vajda V, and Gauthier JA. 2018. Early evolution of modern birds structured by global forest collapse at the end-cretaceous mass extinction. Curr. Biol 28:1825–1831. [DOI] [PubMed] [Google Scholar]
  13. Foth C 2011. The morphology of neoptile feathers: ancestral state reconstruction and its phylogenetic implications. J. Morphol 272:387–403. [DOI] [PubMed] [Google Scholar]
  14. Foth C 2012. On the identification of feather structures in stem-line representatives of birds: evidence from fossils and actuopalaeontology. Paläontol Z. 86:91–102. [Google Scholar]
  15. Foth C 2020. A morphological review of the enigmatic elongated tail feathers of stem birds. Pp. 173–184in Foth C and Rauhut OWM, eds. The evolution of feathers: from their origin to the present. Springer International Publishing, Cham, Swit: zerland. [Google Scholar]
  16. Gill FB, Prum RO, and Robinson SK. 2019. Ornithology. W. H. Freeman and Company, New York. [Google Scholar]
  17. Herte H. 1966. Structure, form, movement. Reinhold Publishing Corporation, New York. [Google Scholar]
  18. Hou L, Zhou Z, Martin LD, and Feduccia A. 1995. A beaked bird from the Jurassic of China. Nature 377:616–618. [Google Scholar]
  19. Jarvis ED, Mirarab S, Aberer AJ, Li B, Houde P, Li C, Ho SYW, Faircloth BC, Nabholz B, Howard JT, et al. 2015. Phylogenomic analyses data of the avian phylogenomics project. GigaScience 4. 10.1186/s13742-014-0038-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Ji Q, Chiappe LM, and Ji S. 1999. A new late Mesozoic confuciusornithid bird from China. J. Vertebr. Paleontol 19:1–7. [Google Scholar]
  21. Jiang T-X, Widelitz RB, Shen W-M, Will P, Wu D-Y, Lin C-M, Jung H-S, and Chuong C-M. 2004. Integument pattern formation involves genetic and epigenetic controls: feather arrays simulated by digital hormone models. Int. J. Dev. Biol 48:117–135. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Lee SH, Fu KK, Hui JN, and Richman JM. 2001. Noggin and retinoic acid transform the identity of avian facial prominences. Nature 414:909–912. [DOI] [PubMed] [Google Scholar]
  23. Lucas AM, and Stettenheim PR. 1972. Avian anatomy: integument. United States Department of Agriculture, Washington, DC. [Google Scholar]
  24. Macleod GD 1980. Mechanical properties of contour feathers. J. Exp. Biol 87:65–71. [Google Scholar]
  25. Mayr G 2009. Paleogene fossil birds. Springer Lond. [Google Scholar]
  26. McCoy VE, Gabbott SE, Penkman K, Collins MJ, Presslee S, Holt J, Grossman H, Wang B, Solórzano Kraemer MM, Delclòs X, et al. 2019. Ancient amino acids from fossil feathers in amber. Sci. Rep 9:6420. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. McKellar RC, Chatterton BDE, Wolfe AP, and Currie PJ. 2011. A diverse assemblage of Late Cretaceous dinosaur and bird feathers from Canadian amber. Science 333:1619–1622. [DOI] [PubMed] [Google Scholar]
  28. Moyer AE, Zheng W, and Schweitzer MH. 2016. Keratin durability has implications for the fossil record: results from a 10 year feather degradation experiment. PLoS One 11:e0157699. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Nesbitt SJ, and Clarke JA. 2016. The anatomy and taxonomy of the exquisitely preserved Green River Formation (Early Eocene) lithornithids (Aves) and the relationships of Lithornithidae. Bull. Am. Mus. Nat. Hist 406:1–91. [Google Scholar]
  30. O’Connor J 2020. The plumage of basal birds. Pp. 147–172 in Foth C and Rauhut OWM, eds. The evolution of feathers: from their origin to the present. Springer International Publishing, Cham, Switzerlan: d. [Google Scholar]
  31. O’Connor JK, Chiappe LM, Chuong C-M, Bottjer DJ, and You H. 2012. Homology and potential cellular and molecular mechanisms for the development of unique feather morphologies in early birds. Geosciences 2:157–177. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Perrichot V, Marion L, Néraudeau D, Vullo R, and Tafforeau P. 2008. The early evolution of feathers: fossil evidence from Cretaceous amber of France. Proc. R. Soc. B: Biol. Sci 275:1197–1202. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Prum RO, and Dyck J. 2003. A hierarchical model of plumage: morphology, development, and evolution. J. Exp. Zool. B Mol. Dev. Evol 298B:73–90. [DOI] [PubMed] [Google Scholar]
  34. Rutschke E 1965. Beiträge zur morphologie der pinguinfeder. Z. Morphol. Öekol. Tiere 55:835–858. [Google Scholar]
  35. Schor R, and Krimm S. 1961. Studies on the structure of feather keratin: II. A β-helix model for the structure of feather keratin. Biophys J. 1:489–515. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Schulte EK 1991. Standardization of biological dyes and stains: pitfalls and possibilities. Histochemistry 95:319–328. [DOI] [PubMed] [Google Scholar]
  37. Shi GH, Grimaldi DA, Harlow GE, Wang J, Wang J, Yang M, Lei W, Li Q, and Li X. 2012. Age constraint on Burmese amber based on U-Pb dating of zircons. Cretaceous Res. 2012:155–163. [Google Scholar]
  38. Stettenheim PR 2000. The integumentary morphology of modern birds—an overview. Am. Zool 40:461–477. [Google Scholar]
  39. Wang B, and Meyers MA. 2017. Light like a feather: a fibrous natural composite with a shape changing from round to square. Adv. Sci 4:1600360. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Wang M, Zheng X, O’Connor JK, Lloyd GT, Wang X, Wang Y, Zhang X, and Zhou Z. 2015. The oldest record of ornithuromorpha from the Early Cretaceous of China. Nat. Commun 6:6987. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Wang X, O’Connor JK, Zheng X, Wang M, Hu H, and Zhou Z. 2014. Insights into the evolution of rachis dominated tail feathers from a new basal enantiornithine (Aves: Ornithothoraces). Biol. J. Linn. Soc 113:805–819. [Google Scholar]
  42. Wu P, Ng CS, Yan J, Lai Y-C, Chen C-K, Lai Y-T, Wu S-M, Chen J-J, Luo W, Widelitz RB, et al. 2015. Topographical mapping of α- and β-keratins on developing chicken skin integuments: functional interaction and evolutionary perspectives. PNAS 112:E6770–E6779. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Xing L, McKellar RC, Wang M, Bai M, O’Connor JK, Benton MJ, Zhang J, Wang Y, Tseng K, Lockley MG, et al. 2016a. Mummified precocial bird wings in mid-Cretaceous Burmese amber. Nat. Commun 7:12089. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Xing L, McKellar RC, Xu X, Li G, Bai M, Persons WS, Miyashita T, Benton MJ, Zhang J, Wolfe AP, et al. 2016b. A feathered dinosaur tail with primitive plumage trapped in mid-Cretaceous amber. Curr. Biol 26:3352–3360. [DOI] [PubMed] [Google Scholar]
  45. Xing L, O’Connor JK, McKellar RC, Chiappe LM, Tseng K, Li G, and Bai M. 2017. A mid-Cretaceous enantiornithine (Aves) hatchling preserved in Burmese amber with unusual plumage. Gondwana Res. 49:264–277. [Google Scholar]
  46. Xing L, Cockx P, McKellar RC, and O’Connor J. 2018. Ornamental feathers in Cretaceous burmese amber: resolving the enigma of rachis-dominated feather structure. J. Palaeogeogr 7:13. [Google Scholar]
  47. Xing L., Cockx P, and McKellar RC. 2020. Disassociated feathers in Burmese amber shed new light on mid-Cretaceous dinosaurs and avifauna. Gondwana Res. 82:241–253. [Google Scholar]
  48. Xu X 2020. Filamentous integuments in nonavialan theropods and their kin: advances and future perspectives for understanding the evolution of feathers. Pp. 67–78 in Foth C and Rauhut OWM, eds. The evolution of feathers: from their origin to the present. Springer International Publishing, Cham, Switzerla: nd. [Google Scholar]
  49. Xu X, and Guo Y. 2009. The origin and earlier evolution of feathers: insights from recent paleontological and neontological data. Vert. PalAs 47:311–329. [Google Scholar]
  50. Yang Z, Jiang B, McNamara ME, Kearns SL, Pittman M, Kaye TG, Orr PJ, Xu X, and Benton MJ. 2019. Pterosaur integumentary structures with complex feather-like branching. Nat. Ecol. Evol 3:24–30. [DOI] [PubMed] [Google Scholar]
  51. Yue Z, Jiang T-X, Widelitz RB, and Chuong C-M. 2006. Wnt3a gradient converts radial to bilateral feather symmetry via topological arrangement of epithelia. PNAS 103:951–955. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Zhang F, and Zhou Z. 2000. A primitive enantiornithine bird and the origin of feathers. Science 290:1955–1959. [DOI] [PubMed] [Google Scholar]
  53. Zhang F, Zhou Z, and Benton MJ. 2008a. A primitive confuciusornithid bird from China and its implications for early avian flight. Sci. China Ser. D: Earth Sci 51:625–639. [Google Scholar]
  54. Zhang F, Zhou Z, Xu X, Wang X, and Sullivan C. 2008b. A bizarre Jurassic maniraptoran from China with elongate ribbon-like feathers. Nature 455:1105–1108. [DOI] [PubMed] [Google Scholar]
  55. Zhao T, Hu J, Hu L, and Pan Y. 2020. Experimental maturation of feathers: implications for interpretations of fossil feathers. Palaios 35:67–76. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Electronic supplementary information text, Figs. S1 to S5, Table S1, and References for the electronic supplementary information.

Table S1. Summary of RDFs preserved in Burmese ambers.

Fig. S1. Photos of RDFs preserved in the early Late Cretaceous Burmese ambers.

Fig. S2. Schematic drawings showing the parameters using in the functional simulations.

Fig. S3. Static stress analyses of feathers with various combinations of vane and rachidial morphologies.

Fig. S4. Tissue differentiation (H&E) of a regenerating contour feather from a one-year old chicken.

Fig. S5. RDFs in extant penguins (Pygoscelis papua).

NIHMS1964876-supplement-Electronic_supplementary_information_text__Figs__S1_to_S5__Table_S1__and_References_for_the_electronic_supplementary_information_.docx (3MB, docx)

Data Availability Statement

Data are available on the Dryad Digital Repository (https://doi.org/10.5061/dryad.18931zctm).

RESOURCES