View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector Current Biology 23, 1065–1074, June 17, 2013 ª2013 Elsevier Ltd All rights reserved http://dx.doi.org/10.1016/j.cub.2013.04.062 Article Developmental Basis of Phallus Reduction during Bird Evolution

Ana M. Herrera,1,4 Simone G. Shuster,2,4 Claire L. Perriton,3,4 Some of the most striking examples of the evolutionary and Martin J. Cohn1,2,3,* diversification of genital form and function are found in birds 1Howard Hughes Medical Institute, Department of Molecular [10–18]. All birds reproduce by internal fertilization, yet only Genetics and Microbiology, and the Genetics Institute w3% of extant birds possess a phallus capable of intromis- 2Department of Biology sion [19]. Indeed, external genitalia are reduced or absent in University of Florida, Gainesville, FL 32610, USA nearly 10,000 species of birds. These evolutionary changes 3School of Animal and Microbial Sciences, University of have resulted in phallus morphologies that can be classified Reading, Reading RG6 6AJ, UK operationally into three general categories: (1) an intromittent phallus that is capable of insertion during copulation (e.g., ducks), (2) a nonintromittent phallus that is reduced but not Summary lost (e.g., chickens), and (3) absence of the phallus that involves a complete loss of a phallic protuberance (e.g., neo- Background: One of the most puzzling events in evolution is aves). Recent progress in phylogenomics has begun to clarify the reduction and loss of the phallus in birds. All birds repro- the phylogenetic relationships among birds [20], and evolu- duce by internal fertilization, but only w3% of birds have tionary reductions of the intromittent phallus can be mapped retained a phallus capable of intromission. A number of to a relatively small number of nodes in the avian phylogeny; hypotheses have been proposed for the evolutionary mecha- for example, after their divergence from Anseriformes (e.g., nisms that drove phallus reduction; however, the underlying ducks, swans, geese, and allies), Galliformes (e.g., chickens, developmental mechanisms are unknown. quails, pheasants, and megapodes) evolved a highly reduced, Results: We investigated genital development in two sister nonintromittent phallic rudiment. Even more extreme reduc- clades of birds, Galliformes (land fowl), most of which lack tion occurred near the base of Neoaves, the largest clade of an intromittent phallus, and Anseriformes (waterfowl), which birds, which have no remnant of the phallus [21, 22]. have well developed phalluses; and in two outgroups, Paleo- Although most galliforms underwent extreme reduction of gnathae (emus) and Crocodilia (alligators). Galliform embryos the external genitalia [21, 22], members of their sister clade, undergo cryptic development of a genital tubercle, the precur- the anseriforms, retained an intromittent phallus, and in sor of the phallus, but this later undergoes apoptosis, leading some species these elongated coiled organs can exceed the to regression of the tubercle. At the molecular level, a derived length of the body [23]. These morphological changes led to pattern of Bmp4 expression was identified in chick (a galliform) changes in copulation strategies in different lineages. For genital tubercles. Inhibition of Bmp signaling in chick genitalia example, male birds that lack a prominent intromittent organ rescues cells from apoptosis and prevents phallus regression, (IO) transfer sperm by apposing the cloaca with that of the whereas activation of Bmp signaling in duck (an anseriform) female, in a maneuver known as the ‘‘cloacal kiss,’’ which genitalia induces a galliform-like pattern of apoptosis. Thus, requires cooperation of the female [10, 11]. By contrast, males distal Bmp activity is necessary and sufficient to induce with an IO can manipulate females and even forcibly copulate apoptosis in Galloanserae genital tubercles. with unwilling females, a behavior that is well documented in Conclusions: Our results indicate that evolutionary reduction waterfowl [10, 13, 24]. of the intromittent phallus in galliform birds occurred not by While selective pressures for elaborate phallus morphol- disruption of outgrowth signals but by de novo activation of ogies are obvious, it is less clear how selection could favor cell death by Bmp4 in the genital tubercle. These findings, reduction or loss of an IO. A number of hypotheses have together with discoveries implicating Bmps in evolution of been proposed for the adaptive significance of genital reduc- beak shape, feathers, and toothlessness, suggest that modu- tion, although the precise nature of the evolutionary mecha- lation of Bmp gene regulation played a major role in the evolu- nisms is unclear due to the paucity of ecological or experi- tion of avian morphology. mental data [10, 11, 19, 21, 22]. One of the most influential hypotheses for genital diversity, the ‘‘lock and key’’ hypothe- Introduction sis, argues that rapid evolution of genitalia facilitated specia- tion through breeding incompatibility (i.e., only males and Few morphological characters in the animal kingdom rival the females of the same species can fit their genitalia together) diversity, complexity, and evolvability of male external geni- [7]. Although anatomical barriers to hybridization can in theory talia [1]. Among animals with internal fertilization, genital lead to reproductive isolation, experimental studies have failed morphology evolves rapidly, and the size, shape, and anatom- to support the idea that strict mechanical incompatibility has ical details of the external genitalia can show dramatic varia- been a driving force in the evolution of genital form [2, 6, 7]. tion, even between closely related species [2]. The basis of Alternatively, Mayr proposed that genital morphology this diversity has been the subject of debate, and a number undergoes frequent changes due to pleiotropic effects [9]. It of evolutionary mechanisms have been proposed to account is unclear why neutral changes resulting from pleiotropy would for the rapid divergence of genital form [1–9]. affect genitalia disproportionately [2], although studies of the genetic control of genital development have identified a multi- tude of developmental mechanisms (including cis-acting DNA 4These authors contributed equally to this work regulatory elements) that are shared by other organ systems *Correspondence: mjcohn@ufl.edu [25–28], which raises the possibility that modulating gene Current Biology Vol 23 No 12 1066

expression in structures such as limbs could have collateral conserved in galliform embryos. To determine when these effects on the genitalia [29] A third hypothesis argues that morphological differences arise, we used scanning electron diversification of external genitalia has been driven by sexual microscopy to investigate genital development in a staged selection acting on variation that affects sperm competition, [44] series of chick and duck embryos. External genital devel- sensory features of genital morphology, female choice, and opment begins at the same embryonic stage (stage 26) in sexually antagonistic coevolution (i.e., males and females chicks and ducks (Figures 1C and 1D). Despite the internal evolve adaptations and counteradaptations to gain control position of the phallus within the cloaca in adult birds, embry- over copulation and fertilization) [2, 6, 7, 12, 30]. onic development of the genital tubercle occurs externally and At a developmental level, little is known about the molecular resembles the mouse genital tubercle at early stages (Figures genetic mechanisms of external genital evolution. Recent 1C–1H) [32] Development of the phallus begins with the initia- progress in the developmental biology of mammalian external tion of paired genital swellings between the anterior and pos- genitalia has led to the identification of genetic pathways that terior cloacal swellings (Figures 1C and 1D). By stage 28, the control outgrowth and patterning of the penis [29, 31–42]. paired swellings have merged to form a single genital tubercle These discoveries, together with independent advances in in both chicks and ducks (Figures 1E–1H). Outgrowth of the avian phylogenetics, reproductive ecology, and comparative tubercle continues until stage 35 in chicks, when development morphology [12, 20, 21, 43], create a new opportunity to arrests, but in ducks the phallus continues to grow and extend investigate the developmental genetic basis of genital evolu- from the cloaca (Figures 1I–1L). In chicks at stage 45, the rudi- tion in birds. Here we report that the same mechanisms that mentary phallus is visible only as a bulge on the inferior margin regulate mammalian penis development also operate in bird of the cloacal collar (Figure 1M). Ducks at the same stage have external genitalia, and that the proliferative cues that direct a pronounced phallus that begins to form the characteristic genital outgrowth are conserved in chick (a galliform) and coiled pattern (Figure 1N). Thus, initiation and early outgrowth duck (an anseriform) embryos despite striking differences in of the genital tubercle occurs in both chicks and ducks, but in their adult genital morphologies. Although early development chickens, outgrowth arrests and the tubercle then regresses of the genital tubercle (the precursor of the phallus) is within the cloaca. conserved in chicks and ducks, we find that chicks have a unique domain of cell death distally. This domain of cell death Mechanisms of Genital Tubercle Outgrowth Are Retained is associated with ectopic activation of the gene encoding in Chicks bone morphogenetic protein 4 (Bmp4), which is known to The precocious arrest of chick phallus development induce apoptosis in other organs. Experimental inactivation suggested that the molecular mechanisms that regulate of Bmp signaling in chick embryos blocks apoptosis and pre- genital tubercle outgrowth are disrupted. In mice, sustained vents regression of the genital tubercle. Furthermore, ectopic outgrowth of the genital tubercle requires Sonic hedgehog application of Bmp protein to the distal tip of duck genital (Shh) and Hoxa13/d13 [28, 32–34, 38, 41, 45]. Given that loss tubercles leads to a chick-like pattern of apoptosis in ducks. of Shh or Hoxa13 and Hoxd13 function in mouse embryos We then extended our study to the most basal group of birds, results in arrest of genital tubercle outgrowth, which mimics the Paleognathae, and to the sister group to the birds, the the situation in chicks, we tested whether these pathways Crocodilia, both of which have intromittent phalluses. The are disrupted in chick genital tubercles. Unexpectedly, strong results show that ducks retained the plesiomorphic pattern expression patterns of Shh, Hoxa13, and Hoxd13 are well of Bmp4 expression and cell death, whereas chickens conserved in chick and duck genitalia, despite the failure of evolved a novel domain of Bmp-mediated cell death that chick genital development to progress beyond the early bud causes regression of the phallus before hatching. Taken stage (Figure 2). Shh is expressed along the ventral sulcus in together, our results suggest that reduction of the intromittent chick and duck at stage 29 and persists through stage 35 phallus in galliform birds evolved not by disruption of (Figures 2A–2D). To determine whether mesenchymal cells outgrowth signals but by de novo activation of cell death in respond to Shh, we examined Ptch1 and Ptch2, transcriptional the developing genital tubercle. targets of Hedgehog [47]. Both genes show strong expression in mesenchyme on either side of the sulcus as late as stage 35 Results (Figures 2E–2L), indicating activation of the Hedgehog pathway. Hoxd13 and Hoxa13 are expressed throughout the Chicks and Ducks Initiate Early Outgrowth of Genital genital tubercles of both chick and duck embryos at stage 29 Tubercles, but Chick Tubercles Arrest and Regress (Figures 2M and 2N, 2Q, and 2R), and expression weakens in Reanalysis of avian external genital morphology in light of both species by stage 35 (Figures 2O and 2P, 2S, and 2T), recent genomic studies of bird interrelationships suggests Thus, despite the extreme reduction of the galliform phallus, that reduction of the phallus likely occurred in stem galliforms the mechanisms known to direct early outgrowth of the genital after their divergence from anseriforms (Figure 1A). Compar- tubercle have been retained in chicks. ison of male genital morphology in two anseriforms, dome- The similar expression patterns of outgrowth-promoting stic graylag goose (Anser anser) and Pekin duck (Anas genes in chick and duck genitalia led us to investigate whether platyrhynchos), and in two galliforms, domestic chicken the precocious arrest of genital development in chicks (Gallus gallus domesticus) and Old World quail (Coturnix involves a failure of genital tubercle cells to respond to prolif- coturnix), reveals striking differences (Figure 1B). Geese and erative cues. Quantitative analysis of cell proliferation in chick ducks have elongated, coiled phalluses with dermal spines, and duck genital tubercle mesenchyme revealed no significant whereas chickens and quails have only rudimentary phallic differences (p = 0.6) in their mitotic indices, despite the swellings on the anterior margin of the cloacal wall (Figure 1B). obvious differences in outgrowth at stage 35 (see Figure S1 Although highly reduced, chicken and quail phallic swellings available online). These results indicate that chick genital retain proximodistal and dorsoventral polarity and a ventral mesenchyme exhibits a normal response to the growth- sulcus, suggesting that early genital patterning may be promoting genes described above. Evolutionary Reduction of the Penis in Birds 1067

Figure 1. Phylogenetic, Adult, and Embryological Comparison of External Genitalia in Birds (A) Phylogenetic distribution of intromittent phalluses in birds. + indicates present; 2 indicates absent or reduced in the majority of species in the clade. *Most tinamous have an intromittent phallus, but an exception has been reported for one species [21]. Phylogeny is from [20]. Branch colors are as follows: purple, Paleognathae; green, Anseriformes; cyan, Galliformes; orange, Neoaves; black, Crocodilia (root). (B) Anatomical preparations of adult male external genitalia in two anseriforms (goose and duck) and two galliforms (chicken and quail). Arrows point to the reduced phalluses in galliform species. Scale bar represents 1 cm. (C–N) Scanning electron microscopy of chick and duck external genital development. Anterior is to the left in (C)–(F) and to the top in (G)–(N). Chick and duck embryos initiate formation of paired genital swellings (C and D), which form a single tubercle during early outgrowth (E and H). Outgrowth arrests and the genital tubercle regresses in chick embryos (I, K, and M), but genital development continues in duck embryos (J, L, and N). Hamburger-Hamilton (HH) stage numbers are shown above. Abbreviations: gs, genital swellings; gt, genital tubercle; acs, anterior cloacal swelling; pcs, posterior cloacal swelling.

Increased Apoptosis Is a Synapomorphic Character chick, whereas goose genitalia at this stage showed an of Galliform Genital Tubercles apoptotic pattern resembling that of duck (Figure S2). By Appendage outgrowth requires a balance between cell prolif- stage 36, quail genital tubercles showed extensive apoptosis eration and cell death, and truncation can result from either throughout the distal mesenchyme and sulcus, whereas goose decreased proliferation or increased cell death [33]. To deter- genital tubercles exhibited very low levels only along the mine whether increased apoptosis could underlie the arrest of sulcus (Figures 3B and 3D). These results indicate that the chick genital development, we compared patterns of cell galliforms examined here (chicken and quail) undergo a late death in the genital tubercles of galliform, anseriform, paleo- wave of apoptosis in the distal genital tubercle, which coin- gnath, and crocodilian embryos. LysoTracker Red staining cides temporally with the arrest of outgrowth and onset of shows that chicks and ducks have similar patterns of regression. apoptosis at stage 29, with labeled cells along the sulcus in If the evolutionary reduction of galliform external genitalia both species (Figure S2), in a pattern similar to that reported was driven by increased apoptosis, then the pattern of cell for E11.5 mice [32]. Comparison at stage 36, however, death observed in chick and quail should be a synapomorphic revealed marked differences: chick genital tubercles showed (shared derived) character, whereas the duck and goose widespread apoptosis throughout the distal mesenchyme pattern should be plesiomorphic (ancestral). We tested this and along the sulcus, whereas duck genitalia had few labeled hypothesis by extending our study to the most basal clade cells in the sulcus or distal region of the tubercle (Figures 3A of birds, the Palaeognathae, and to the sister group to birds, and 3C). To determine whether the differences in apoptosis the Crocodilia. Analysis of the emu, a paleognath, showed between chick and duck genital tubercles are conserved fea- that there is low-level cell death along the sulcus at stage 36, tures of Galliformes and Anseriformes, respectively, we exam- as seen in ducks and geese (Figure 3E). In comparably staged ined additional members of each clade. Quail genital tubercles alligator embryos, the genital tubercles show even fewer at stage 29 showed an apoptotic pattern similar to that seen in apoptotic cells than in emus, ducks, and geese (Figure 3F). Current Biology Vol 23 No 12 1068

(Figures 3I0–3L0). To quantify levels of Bmp4 expression in chick and duck genital tubercles, we used quantitative real- time PCR (qRT-PCR) and found that chick genitalia have significantly higher levels of Bmp4 mRNA at stage 29 (>3-fold higher) and stage 35 (>2-fold higher; Figure 3Q). The sustained expression of Bmp4 in chick genitalia led us to ask whether Msx1 and Msx2, downstream effectors of Bmp signaling [48], show differential activity in chick and duck genital tubercles. At stages 29 and 35, Msx1 and Msx2 were expressed distally in chick genitalia (Figures 3M, 3M0, 3O, and 3O0), where the region of apoptosis occurs (Figure 3A), but neither gene could be detected in duck genitalia (Figures 3N, 3N0, 3P, and 3P0). These findings show that distal activation of the Bmp4 pathway coincides spatially and temporally with the wave of cell death that occurs in galliform external genitalia. The results described above raised the possibility that galli- forms evolved a new domain of Bmp4 (and apoptosis) in their distal genital tubercles, but comparison of duck and chick genital development could not exclude the possibility that distal Bmp4 expression is a primitive character that was lost in anseriforms. Therefore, we examined Bmp4 expression in the emu, a basal bird with a well-developed phallus that does not undergo widespread distal apoptosis (Figure 3E). In Figure 2. Comparative Analysis of Gene Expression in Developing External emu embryos at stage 29, Bmp4 showed a duck-like pattern Genitalia of Chick and Duck Embryos of expression: Bmp4 expression was detected in the cloacal Whole-mount RNA in situ hybridizations of chick and duck genital tubercles collar and interdigitally in the hindlimbs, but not at the distal at stages 29 and 35 showing expression patterns of genes that promote tip of the genital tubercle (Figures 3R and 3S). Taken together mouse external genital development [28, 29, 32, 34, 45, 46]. Genital tuber- with our phylogenetic survey of apoptosis across archosaur cles are shown in ventral view with distal toward the top. Gene names are (avian + crocodilian) external genitalia (Figures 3A–3F), these shown at top; species and Hamburger-Hamilton (HH) stage are shown at left. results indicate that Bmp4 expression and apoptosis in the distal genital tubercle evolved in the galliform lineage after its divergence from the anseriforms. Thus, the wave of cell death in the distal genital tubercle appears to be a synapomorphy of galliforms. Inhibition of Bmp Signaling Rescues Apoptosis and Prevents Regression of Chick Genital Tubercles Chicken Genital Tubercles Have a Derived Pattern of Bmp4 Having established the relative timing of the evolutionary Expression acquisition of distal Bmp4 expression, we went on to perform The finding that regression of galliform genital tubercles is functional experiments to test the hypothesis that cell death in associated with apoptosis, rather than loss of an outgrowth galliform external genitalia is caused by Bmp signaling. If this signal, raised new questions about the molecular mechanism hypothesis is correct, then antagonism of Bmp activity should responsible for the distal cell death. Bone morphogenetic pro- rescue cells from apoptosis. We applied beads soaked in teins (Bmps) have been shown to induce apoptosis in a num- Noggin protein (0.275 mg/ml), which binds Bmps and blocks ber of developmental contexts, including the external genitalia their interaction with Bmp receptors [51], to one side of chick of mice [31, 48–50]. To determine whether endogenous differ- genital tubercles at stage 26. Noggin beads resulted in local ences in Bmp activity could underlie the differences in inhibition of apoptosis within 24 hr, whereas the contralateral apoptosis between galliform and anseriform genitalia, we side showed no change in apoptosis (Figure 4A). Chick tuber- compared expression of Bmp2, Bmp4, and Bmp7 (ligands cles receiving control beads (soaked in PBS) showed no for BmpR1a, which controls apoptosis of the mouse genital change in apoptosis on either side (Figure 4B). To confirm tubercle [31]) in chick and duck embryos. Chicks and ducks the specificity of Noggin treatments, we monitored expression showed different patterns of Bmp4 and Bmp7 expression at of the Bmp target gene Msx1. On the Noggin-treated side of stage 29 (Figures 3G–3P), before differences in phallus devel- the genital tubercle, Msx1 was downregulated within 24 hr opment can be detected (Figures 1I and 1J). In chicks, strong (the contralateral side, by contrast, continued to show strong expression of Bmp4 was detected along the entire proximo- expression of Msx1), indicating that Bmp signaling had been distal axis of the genital tubercle (Figure 3G), whereas in inactivated (Figure 4C). PBS control beads did not affect duck genitalia, Bmp4 was expressed in the cloaca and at the Msx1 expression (Figure 4D). base of the genital tubercle, but not at the distal tip (Figure 3H). Interestingly, the distal tip of the genital tubercle appeared Bmp7 also showed strong expression along the chick genital to extend further on the Noggin-treated side than on the tubercle at stage 29, particularly at the distal tip, but compar- contralateral side (Figure 4C), suggesting that increased cell atively little expression was seen in duck genitalia (Figures 3I survival resulted in sustained outgrowth. We therefore quanti- and 3J). fied proximodistal outgrowth of the bead side versus the At stage 35, Bmp4 continued to be expressed distally in contralateral side 24 hr after implantation of either a Noggin chicks only (Figures 3G0 and 3H0). By this stage, Bmp7 and bead or PBS control bead. The difference between the bead Bmp2 expression had diminished in both chick and ducks side and contralateral side was w6.5 times greater in Evolutionary Reduction of the Penis in Birds 1069

Figure 3. Apoptosis and Bmp Signaling Are Increased in Galliform Genital Tubercles (A–F) LysoTracker Red staining shows widespread apoptosis in galliform (chick and quail) genital tubercles, whereas anseriform (duck and goose), emu, and alligator genitalia at comparable stages show limited apoptosis mostly along the sulcus. Phylogenetic relationships [20] are shown at top. (G–P) Bmp4, Bmp7, Bmp2, Msx1, and Msx2 expression in chicks and ducks at stage 29 (G–P) and 35 (G0–P0). Bmp4 (G and H), Bmp7 (I and J), Msx1 (M and N), and Msx2 (O and P) show distal expression in chick but not duck genital tubercles at stage 29 (G–J). Distal expression domains of Bmp4 (G0 and H0), Msx1 (M0 and N0), and Msx2 (O0 and P0) persist in chick genital tubercles through stage 35. Bmp2 shows no significant difference at either stage (K, L, K0, and L0). (Q) qRT-PCR shows that Bmp4 expression is significantly higher in genital tubercles (GTs) of chicks relative to ducks at stage 29 (p = 0.002) and stage 35 (p = 0.0003). Error bars show standard error of the mean. (R) In the emu genital tubercle, Bmp4 expression is absent from the distal tip. Expression is limited to the base of the tubercle and the cloacal collar, as in ducks (compare with H). (S) Internal control of emu hindlimb showing interdigital expression of Bmp4.

Noggin-treated tubercles (100.8 mm) than in control tubercles the genital tubercles of duck embryos at stage 26, the stage (15.6 mm; p = 0.01; Figure 4E). Thus, antagonism of Bmp when Bmp is required for apoptosis in chicks. Eight hours after signaling in the chick genital tubercle resulted in significantly application of a Bmp2 bead, duck genital tubercles showed increased outgrowth. These results indicate that Bmp function ectopic apoptosis on the treated side of the tubercle, whereas is required for apoptosis and the arrest of outgrowth in chick application of control beads had no effect on cell death (Fig- genital tubercles. ure S3). Taken together, these results show that distal Bmp The results described above led us to ask whether, during activity is both necessary and sufficient for apoptosis and the evolution of Galloanserae, activation of Bmp signaling regression of the phallus in Galloanserae. The findings pre- could have been sufficient to induce apoptosis in a genital sented here suggest that a novel domain of Bmp4 arose in tubercle fated to form an intromittent phallus. We tested this the distal genital tubercle during the evolution of galliforms, hypothesis by applying beads loaded with Bmp2 protein and this led to apoptosis-meditated regression of the phallus (a paralog of Bmp4 that also signals through BmpR1a [52]) to (Figure 4F). Current Biology Vol 23 No 12 1070

Figure 4. Bmp Signaling Causes Apoptosis and Inhibition of Outgrowth in Chick Genitalia (A–D) Beads were loaded with either Noggin protein (A and C) or PBS (B and D) and implanted into the left side (right in photos) of stage 29 chick genital tubercles. (A) Noggin bead (arrow) decreases apoptosis (red cells are LysoTracker positive) on the treated side of tubercles. Contralateral side shows widespread apoptosis. (B) PBS bead has no effect on either side. (C) Msx1 is downregulated on the side of the tubercle with Noggin bead (circle), but expression is maintained on the contralateral side. Note that the Noggin- treated side shows more outgrowth. (D) Msx1 is expressed in a normal pattern on both sides of the tubercle receiving PBS bead (circle). (E) Comparison of outgrowth in treated versus contralateral sides in Noggin-treated (green) and PBS control-treated (red) genital tubercles 24 hr after bead implantation. Noggin beads induced a mean increase of 100.8 mm over the contralateral side, which is significantly greater than the difference between PBS control and contralateral sides (15.8 mm; p = 0.01). (F) Developmental model for evolutionary loss of the intromittent phallus in galliforms. Black circles indicate presence of an intromittent phallus; open circle with diagonal line indicates loss of intromittent phallus; asterisk indicates all galliforms with the exception of cracids, which have a fully intromittent phallus.

Discussion galliforms. The results presented above suggest that a new region of Bmp4 expression arose in the galliform lineage after During the evolution of birds, an intromittent phallus was its divergence from anseriforms, and that this resulted in a reduced or lost in multiple lineages, including at least once domain of apoptosis that arrests genital outgrowth and causes in galliforms and again in the common ancestor of neoaves regression of the phallus. [19, 21, 22]. Consequently, w97% of extant bird species lack an intromittent phallus [10, 11, 22]. Our comparative study of Evolution of Phallus Reduction external genital development in birds shows conservation of The marked divergence of genital morphology between the signals that promote initiation and early outgrowth of the anseriforms and galliforms raises questions about the selec- genital tubercle across galliform, anseriform, and paleognath tive pressures that led to phallic reduction. Loss of intromittent lineages, but our data from chick embryos suggest that galli- organs is an evolutionary paradox: How can reduction of a forms are unique in their expression of Bmp4 in the distal gen- structure that facilitates internal fertilization have a positive ital tubercle. Experimental manipulations of Bmp signaling in effect on reproductive fitness? It is unlikely that reduction of chick and duck embryos show that this derived domain of the phallus resulted in improved delivery of sperm. Darwin first Bmp activity is necessary and sufficient for the wave of distal proposed that variable characters with no apparent adaptive apoptosis that causes regression of the genital tubercle in value were not subject to natural selection, but that these traits Evolutionary Reduction of the Penis in Birds 1071

could be stabilized in a population if they confer an advantage conservation between galliforms and anseriforms, suggesting ‘‘over other individuals of the same sex and species, in exclu- that galliforms have not lost the mechanisms for external gen- sive relation to reproduction’’ [53]. Copulation with males lack- ital development. Indeed, quantification of cell division in chick ing an intromittent phallus requires female cooperation, such and duck genital tubercles showed that there is no significant as presentation of the cloaca and eversion of the vagina [10, difference in cell proliferation. This is consistent with our pre- 11]. Accordingly, by selecting males with reduced or nonintro- vious report that, in the mouse, Shh promotes outgrowth of mittent phalluses, females can control paternity, which is the genital tubercle by regulating cell-cycle kinetics [33]. By consistent with the idea that female choice played a role in contrast, galliforms have a derived pattern of apoptosis at reduction and loss of the phallus [10, 11, 14, 19, 22, 43]. the distal tip of the genital tubercle, and this coincides with a An alternative to the sexual selection hypothesis is that domain of Bmp4 expression not detected in anseriforms or reduction of phallus size may have been a secondary or pleio- paleognaths. These results indicate that the distal expression tropic effect of developmental changes to other characters [9]. of Bmp4 and the resulting increase in apoptosis in the distal The molecular mechanisms of appendage development have region of the genital tubercle are a synapomorphic character- been deeply conserved during animal evolution, and the istic of galliforms. gene networks that regulate development of external genitalia The ability of the Bmp antagonist Noggin to prevent cell also control formation of a wide range of other structures, death and to sustain outgrowth of galliform genital tubercles including limbs, gut, nervous system, muscle, and integumen- provides a causal link between the distal domain of Bmp tary appendages such as feathers and scales. Moreover, tis- signaling and apoptotic regression of the phallus. Furthermore, sue-specific expression of a number of genes expressed in our finding that a distal wave of apoptosis can be induced by the genital tubercle, including Bmp4, Msx2, and Hoxd13, has experimental activation of Bmp signaling at the distal tip of been shown to be controlled by enhancers that also direct duck genital tubercles demonstrates that Bmp activity alone transcription in other regions of the embryo. In mouse, an is sufficient to induce cell death and regression of a genital evolutionarily conserved enhancer located more than 45 kb tubercle that otherwise is fated to form an elongated phallus. 50 to the Bmp4 promoter can drive expression in the genital Thus, Bmp signaling in the distal genital tubercle of Galloan- tubercle, digit tips, dorsal root ganglia, and whisker hairs serae is both necessary and sufficient to trigger regression of [25]. Similarly, Msx2 expression in the genital tubercle ecto- the phallus during embryonic development. These findings derm and in the apical ectodermal ridge of the limb buds is are consistent with studies of Bmp function in mouse external regulated by a shared cis-regulatory element [27], as is genital development. Deletion of Bmpr1A results in reduced Hoxd13 expression in the digits and the genital tubercle [54]. cell death in the mouse genital tubercle, which leads to hyper- In the latter example, the length of the digits and the penis plasia of the penis [31]. Moreover, Noggin knockout mice have scale together in response to changes in the quantity of excessive Bmp signaling, which causes reduction of the penis Hoxd13 expression [28]. Therefore, modulation of gene regula- [31]. Taken together, our findings support the hypothesis that tion in other organ systems has the potential to have pleio- evolutionary reduction of an intromittent phallus in galliform tropic effects on gene expression in the genital tubercle (and birds resulted from the acquisition of a new distal domain of vice versa). However, even if reduction of the intromittent Bmp activity in the developing genital tubercle. phallus arose due to pleiotropic effects of Bmp4 regulatory Interestingly, cracids are an exception to the galliform rule in evolution in another cell population and therefore was not that they have a fully intromittent phallus (although not as pro- the primary target of selection, persistence of a reduced nounced as in anseriforms [21]). In light of current phylogeny phallus in galliforms would have required stabilizing selection. [20], cracids may have re-evolved an intromittent phallus Thus, we propose that regression of the intromittent phallus (Figure 1A), which would suggest that galliforms have retained in galliforms arose due to acquisition of a novel domain of the competence to redevelop a phallus from a rudimentary Bmp4 expression in the distal region of the genital tubercle, genital organ. Our findings that (1) the molecular machinery and that stabilization of this domain may have been due to for genital tubercle outgrowth has not been disrupted in selection for males with small or nonintromittent phalluses. chicks, (2) regression of the phallus is caused by Bmp- mediated apoptosis, and (3) antagonism of Bmp signaling is A Developmental Mechanism for Reduction of the Phallus sufficient to block apoptosis and rescue outgrowth raise the in Galliform Birds possibility that modulating Bmp activity may have been Outgrowth of appendages generally requires induction and sufficient for reemergence of an intromittent phallus. maintenance of cell proliferation as well as attenuation of Whether the same developmental mechanism that we apoptosis in order to keep cell death below the rate of cell identified in galliforms also played a role in the independent proliferation. Regionalized regulation of cell death plays an reductions of the phallus in neoaves remains to be tested. important role in morphological sculpting of appendages, Comparative studies of birds have the potential to identify influencing processes and traits such as digit differentiation, the developmental mechanisms responsible for a number of joint cavitation, removal of interdigital webbing, urethra forma- other major morphological transitions, such as how genital tion, and limb and genital length. Activation of apoptosis by outgrowth is sustained for such an extensive period in water- Bmp signaling has been implicated in a number of evolutionary fowl, how coiling of the penis and vagina is regulated, what reductions, such as loss of teeth in birds [55] and reduction of determines the direction of the coiling, and how the micro- forewings in soldier ants [56]. Reciprocally, inhibition of the anatomy of penile spines and other dermal ornamentation is apoptotic activity of Bmps through upregulation of Bmp specified in birds. antagonists (e.g., Gremlin) and/or survival cues (e.g., Fgf8) has been implicated in other evolutionary innovations, such Bmp Regulation and the Evolution of Bird Morphology as retention of interdigital webbing in bat wings. Our analysis Modulation of Bmp expression has been implicated in evolu- of signals known to promote outgrowth of the mouse genital tion of at least four morphological innovations in birds: tubercle, including Shh and Hoxd13/a13, revealed surprising feathers [57], toothlessness [55], beak shape [58, 59], and, Current Biology Vol 23 No 12 1072

now, phallic reduction. It is interesting that the phallus is one of concentration of 1 mg/ml (a gift from Vicki Rosen, Harvard School of Dental several avian structures, including teeth and the most anterior Medicine), or PBS (control) for a minimum of 1 hr. Beads were implanted and posterior digits, that undergo partial development during under the surface ectoderm on one side of the distal tubercle via an incision made with a tungsten needle. Duck genitalia were manipulated in ovo and embryogenesis and then regress. It is possible that this also collected for LysoTracker staining after 8 hr. Chick genital tubercles were reflects gene regulatory elements that are shared among mul- explanted to organ culture in Dulbecco’s modified Eagle’s medium with tiple organ systems. If transcription of a gene in a developing L-glutamine (Invitrogen) with 10% fetal calf serum and 13 Antibiotic-Anti- organ is regulated by an enhancer that also regulates gene mycotic (Gibco/Invitrogen) for 24 hr and then fixed in 4% PFA at 4C for expression in the progenitors of an organ that has been lost further analysis. Proximodistal outgrowth of the genital tubercle was during evolution, then that early phase of expression (and measured using NIH ImageJ. perhaps an early phase of development) could be maintained Analysis of Cell Proliferation in both cell populations. Such conservation of gene regulatory Chick and duck embryos were incubated to stage 35. Embryos (n = 7 for mechanisms may underlie the occasional reemergence of each species) were then harvested, fixed in 4% PFA, and processed for his- structures that have been lost during evolution. tology. Coronal sections (16 mm) were stained with mouse anti-PCNA anti- body (Abcam) at a concentration of 1:1,000 and developed using an ABC Experimental Procedures Kit Elite (Vector Labs) according to the manufacturer’s instructions. Cell counts were made by positioning a counting box (6,670 3 8,940 mm) at Embryo Collection the distal tip of the genital tubercle, immediately under the distal ectoderm. Fertilized chicken, quail, duck, goose, and emu eggs were obtained from Three sections were counted for each embryo, one at the midpoint of the commercial suppliers, and alligator eggs were obtained from the Rockefel- tubercle (the longest proximal-to-distal section) and one on either side ler National Wildlife Refuge. Chicken, quail, goose, and duck eggs were (anterior and posterior) of the midpoint. The three sections were used to incubated in a humidified incubator at 38C, and emu eggs were incubated calculate a mean for each embryo. Significance was determined by ANOVA. at 36.4C. Stages of development were determined using the Hamburger- Hamilton staging series for comparison of multiple anatomical landmarks Supplemental Information [44]. Alligator eggs were incubated at 33C to induce male development. Supplemental Information includes three figures and can be found with this Scanning Electron Microscopy article online at http://dx.doi.org/10.1016/j.cub.2013.04.062. Genital tubercles were dissected in PBS, fixed in 1% glutaraldehyde in PBS, and stored at 4C. After postfixation in 1% osmium tetroxide, specimens Acknowledgments were dehydrated through a graded ethanol series and placed in acetone. Specimens were then critical-point dried, mounted on metal studs, sputter We thank Rebecca Kimball, Ehab Abouheif, Brian Harfe, and members of coated with gold particles, and imaged on a Hitachi S4000 scanning elec- our laboratory for discussions; Patricia Brennan and two other anonymous tron microscope. reviewers for critical comments on the manuscript; Richard Harland and Vicki Rosen for generous gifts of Noggin and Bmp2 protein, respectively; RNA Isolation, cDNA Synthesis, and RT-PCR Patsy Fitch-Hand, Ruth Elsey, and Mark and Cheryl Wagner for eggs and Genital tubercles were dissected from staged embryos and stored in cadavers; and Krista McCoy for assistance with quantitative analysis. This RNAlater. Tissue samples were collected from each embryo for genomic research was supported by a grant from the National Science Foundation DNA extraction and genotyping using sex-specific primers [60]. Male (IOS-0843590). tubercles were pooled and RNA was extracted using an RNeasy Mini Kit (QIAGEN). RNA quantity and purity were determined using a NanoDrop Received: March 13, 2013 ND-1000. For each pooled sample, cDNA was synthesized from 2 mg of total Revised: April 18, 2013 RNA according to manufacturer’s specifications (Bio-Rad). Total RT-PCR Accepted: April 23, 2013 reactions (10 ml) were performed using SYBR Green qPCR Master Published: June 6, 2013 Mix (Bio-Rad). The primers used for RT-PCR were as follows: Bmp4 0 0 0 (F 5 -AACTCCACCAACCACGCCATC-3 ,R5-AGCACCACCTTGTCATACT References CAT-30), chick b-actin (F 50-TGATGGACTCTGGTGATGGTGTTAC-30,R50- CTCTCGGCTGTGGTGGTGAAG-30), duck b-actin (F 50-GCCTCCCTGTC 1. Eberhard, W.G. (2009). Postcopulatory sexual selection: Darwin’s omis- CACCTTCC-30,R50-GCTGCTGATACCTTCACCATTCC-30). Relative level sion and its consequences. Proc. Natl. Acad. Sci. USA 106(Suppl 1 ), of expression was calculated using the DDCt method. 10025–10032. 2. Hosken, D.J., and Stockley, P. (2004). Sexual selection and genital evo- In Situ Hybridization lution. Trends Ecol. Evol. 19, 87–93. Embryos harvested for in situ hybridization were fixed in 4% paraformalde- 3. Eberhard, W.G., and Cordero, C. (1995). Sexual selection by cryptic hyde (PFA) at 4C overnight and dehydrated in a graded methanol series. female choice on male seminal products - a new bridge between sexual Whole-mount RNA in situ hybridization was performed as described by selection and reproductive physiology. Trends Ecol. Evol. 10, 493–496. Nieto et al. [61] with the following modifications: Proteinase K concentration 4. Arnqvist, G., and Thornhill, R. (1998). Evolution of animal genitalia- pat- was increased to 70 mg/ml, following the protocol of Laufer et al. [62]; KTBT terns of phenotypic and genotypic variation and condition dependence buffer contained 1% Triton X-100; alkaline phosphatase buffer (NTMT) of genital and non-genital morphology in water strider. Genet. Res. contained 0.1% Tween-20; and dimethylformamide (10%) was added to Camb. 71, 193–212. NBT-BCIP staining solution. 5. Martin, O.Y., and Hosken, D.J. (2003). The evolution of reproductive isolation through sexual conflict. Nature 423, 979–982. Detection of Apoptotic Cells 6. Eberhardt, W.G. (1985). Sexual Selection and Animal Genitalia Apoptotic cells were localized using LysoTracker Red (Molecular Probes/ (Cambridge: Harvard University Press). Invitrogen). Embryos were rinsed briefly in PBS and incubated in 7. Masly, J.P. (2012). 170 years of ‘‘lock-and-key’’: genital morphology and LysoTracker Red (1:200) in PBS at 37C for 30 min and then washed briefly reproductive isolation. Int. J. Evol. Biol. 2012, 247352. in PBS before fixation in 4% PFA. Samples were rinsed in methanol to 8. Eberhard, W.G. (2010). Evolution of genitalia: theories, evidence, and remove background, viewed and photographed under epifluorescence, new directions. Genetica 138, 5–18. and then photographed again in the same plane of focus with bright-field 9. Mayr, E. (1963). Animal Species and Evolution (Cambridge: Harvard illumination. University Press). 10. Briskie, J.V., and Montgomerie, R. (1997). Sexual selection and the intro- Bead Implantation mittent organ of birds. J. Avian Biol. 28, 73–86. Heparin acrylic beads (Sigma) were size selected (125–250 mm) and soaked 11. Briskie, J.V., and Montgomerie, R. (2001). Efficient copulation and the in either Xenopus Noggin protein (a gift from Richard Harland, University of evolutionary loss of the avian intromittent organ. J. Avian Biol. 32, California, Berkeley) at a concentration of 0.275 mg/ml, Bmp2 protein at a 184–187. Evolutionary Reduction of the Penis in Birds 1073

12. Brennan, P.L., Prum, R.O., McCracken, K.G., Sorenson, M.D., Wilson, 36. Seifert, A.W., Bouldin, C.M., Choi, K.S., Harfe, B.D., and Cohn, M.J. R.E., and Birkhead, T.R. (2007). Coevolution of male and female genital (2009). Multiphasic and tissue-specific roles of sonic hedgehog in morphology in waterfowl. PLoS ONE 2, e418. cloacal septation and external genitalia development. Development 13. Brennan, P.L., Clark, C.J., and Prum, R.O. (2010). Explosive eversion 136, 3949–3957. and functional morphology of the duck penis supports sexual conflict 37. Seifert, A.W., Yamaguchi, T.P., and Cohn, M.J. (2009). Functional and in waterfowl genitalia. Proc. Biol. Sci. 277, 1309–1314. phylogenetic analysis shows that Fgf8 is a marker of genital induction 14. Briskie, J. (1998). Avian genitalia. Auk 15, 826–829. in mammals but is not required for external genital development. 15. Briskie, J.R., Montgomerie, R., and Birkhead, T.R. (1997). The evolution Development 136, 2643–2651. of sperm size in birds. Evolution 51, 937–945. 38. Miyagawa, S., Moon, A., Haraguchi, R., Inoue, C., Harada, M., Nakahara, 16. Birkhead, T.R., and Moller, A.P. (1992). Sperm Competition in Birds: C., Suzuki, K., Matsumaru, D., Kaneko, T., Matsuo, I., et al. (2009). Evolutionary Causes and Consequence (London: Academic Press). Dosage-dependent hedgehog signals integrated with Wnt/beta-catenin 17. King, A.S. (1981). Phallus. In Form and Function in Birds, Volume 2, A.S. signaling regulate external genitalia formation as an appendicular pro- King and J. McLelland, eds. (London: Academic Press), pp. 107–147. gram. Development 136, 3969–3978. 18. Lake, P.E. (1981). Male genital organs. In Form and Function in Birds, 39. Lin, C., Yin, Y., Long, F., and Ma, L. (2008). Tissue-specific requirements Volume 2, A.S. King and J. McLelland, eds. (London: Academic of beta-catenin in external genitalia development. Development 135, Press), pp. 1–61. 2815–2825. 19. Montgomerie, R. (2010). Sexual conflict and the intromittent organs of 40. Li, J., Willingham, E., and Baskin, L.S. (2006). Gene expression profiles male birds. In The Evolution of Primary Sexual Characters in Animals, in mouse urethral development. BJU Int. 98, 880–885. J.L. Leonard and A. Co´ rdoba-Aguilar, eds. (Oxford: Oxford University 41. Lin, C., Yin, Y., Veith, G.M., Fisher, A.V., Long, F., and Ma, L. (2009). Press), pp. 453–470. Temporal and spatial dissection of Shh signaling in genital tubercle 20. Hackett, S.J., Kimball, R.T., Reddy, S., Bowie, R.C., Braun, E.L., Braun, development. Development 136, 3959–3967. M.J., Chojnowski, J.L., Cox, W.A., Han, K.L., Harshman, J., et al. (2008). 42. Petiot, A., Perriton, C.L., Dickson, C., and Cohn, M.J. (2005). A phylogenomic study of birds reveals their evolutionary history. Development of the mammalian urethra is controlled by Fgfr2-IIIb. Science 320, 1763–1768. Development 132, 2441–2450. 21. Brennan, P.L.R., Birkhead, T.R., Zyskowski, K., Waag, J.V.D., and Prum, 43. Brennan, P.L., and Prum, R.O. (2012). The limits of sexual conflict in the R.O. (2008). Independent evolutionary reductions of the phallus in basal narrow sense: new insights from waterfowl biology. Philos. Trans. R. birds. J. Avian Biol. 39, 487–492. Soc. Lond. B Biol. Sci. 367, 2324–2338. 22. Montgomerie, R., and Briskie, J.V. (2007). Anatomy and evolution of 44. Hamburger, V., and Hamilton, H.L. (1951). A series of normal stages in copulatory structures. In Reproductive Biology and Phylogeny of the development of the chick embryo. J. Morphol. 88, 49. Birds, Part A: Phylogeny, Morphology, Hormones and Fertilization, 45. Morgan, E.A., Nguyen, S.B., Scott, V., and Stadler, H.S. (2003). Loss of B.G.M. Jamieson, ed. (Enfield: Science Publishers), pp. 115–148. Bmp7 and Fgf8 signaling in Hoxa13-mutant mice causes hypospadia. 23. McCracken, K.G., Wilson, R.E., McCracken, P.J., and Johnson, K.P. Development 130, 3095–3109. (2001). Sexual selection. Are ducks impressed by drakes’ display? 46. Yamaguchi, T.P., Bradley, A., McMahon, A.P., and Jones, S. (1999). A Nature 413, 128–129. Wnt5a pathway underlies outgrowth of multiple structures in the verte- 24. Brennan, P.L., and Prum, R.O. (2011). The erection mechanism of the brate embryo. Development 126, 1211–1223. ratite penis. J. Zool. 286, 140–144. 47. Pearse, R.V., 2nd, Vogan, K.J., and Tabin, C.J. (2001). Ptc1 and Ptc2 25. Chandler, K.J., Chandler, R.L., and Mortlock, D.P. (2009). Identification transcripts provide distinct readouts of Hedgehog signaling activity of an ancient Bmp4 mesoderm enhancer located 46 kb from the pro- during chick embryogenesis. Dev. Biol. 239, 15–29. moter. Dev. Biol. 327, 590–602. 48. Marazzi, G., Wang, Y., and Sassoon, D. (1997). Msx2 is a transcriptional 26. Spitz, F., Gonzalez, F., and Duboule, D. (2003). A global control region regulator in the BMP4-mediated programmed cell death pathway. Dev. defines a chromosomal regulatory landscape containing the HoxD clus- Biol. 186, 127–138. ter. Cell 113, 405–417. 49. Merino, R., Rodriguez-Leon, J., Macias, D., Gan˜ an, Y., Economides, 27. Sumoy, L., Wang, C.K., Lichtler, A.C., Pierro, L.J., Kosher, R.A., and A.N., and Hurle, J.M. (1999). The BMP antagonist Gremlin regulates Upholt, W.B. (1995). Identification of a spatially specific enhancer outgrowth, chondrogenesis and programmed cell death in the devel- element in the chicken Msx-2 gene that regulates its expression in the oping limb. Development 126, 5515–5522. apical ectodermal ridge of the developing limb buds of transgenic 50. Ferrari, D., Lichtler, A.C., Pan, Z.Z., Dealy, C.N., Upholt, W.B., and mice. Dev. Biol. 170, 230–242. Kosher, R.A. (1998). Ectopic expression of Msx-2 in posterior limb 28. Kondo, T., Za´ ka´ ny, J., Innis, J.W., and Duboule, D. (1997). Of fingers, bud mesoderm impairs limb morphogenesis while inducing BMP-4 toes and penises. Nature 390, 29. expression, inhibiting cell proliferation, and promoting apoptosis. Dev. 29. Cohn, M.J. (2011). Development of the external genitalia: conserved and Biol. 197, 12–24. divergent mechanisms of appendage patterning. Dev. Dyn. 240, 1108– 51. Zimmerman, L.B., De Jesu´ s-Escobar, J.M., and Harland, R.M. (1996). 1115. The Spemann organizer signal noggin binds and inactivates bone 30. Arnqvist, G., and Danielsson, I. (1999). Copulatory behavior, genital morphogenetic protein 4. Cell 86, 599–606. morphology, and male fertilization success in water striders. Evolution 52. Chen, D., Ji, X., Harris, M.A., Feng, J.Q., Karsenty, G., Celeste, A.J., 53, 147–156. Rosen, V., Mundy, G.R., and Harris, S.E. (1998). Differential roles for 31. Suzuki, K., Bachiller, D., Chen, Y.P., Kamikawa, M., Ogi, H., Haraguchi, bone morphogenetic protein (BMP) receptor type IB and IA in differen- R., Ogino, Y., Minami, Y., Mishina, Y., Ahn, K., et al. (2003). Regulation of tiation and specification of mesenchymal precursor cells to osteoblast outgrowth and apoptosis for the terminal appendage: external genitalia and adipocyte lineages. J. Cell Biol. 142, 295–305. development by concerted actions of BMP signaling [corrected]. 53. Darwin, C. (1871). The Descent of Man and Selection in Relation to Sex Development 130, 6209–6220. (London: John Murray). 32. Perriton, C.L., Powles, N., Chiang, C., Maconochie, M.K., and Cohn, 54. Montavon, T., Le Garrec, J.F., Kerszberg, M., and Duboule, D. (2008). M.J. (2002). Sonic hedgehog signaling from the urethral epithelium con- Modeling Hox gene regulation in digits: reverse collinearity and the trols external genital development. Dev. Biol. 247, 26–46. molecular origin of thumbness. Genes Dev. 22, 346–359. 33. Seifert, A.W., Zheng, Z., Ormerod, B.K., and Cohn, M.J. (2010). Sonic 55. Chen, Y., Zhang, Y., Jiang, T.X., Barlow, A.J., St Amand, T.R., Hu, Y., hedgehog controls growth of external genitalia by regulating cell cycle Heaney, S., Francis-West, P., Chuong, C.M., and Maas, R. (2000). kinetics. Nat. Commun. 1, 23. Conservation of early odontogenic signaling pathways in Aves. Proc. 34. Haraguchi, R., Mo, R., Hui, C., Motoyama, J., Makino, S., Shiroishi, T., Natl. Acad. Sci. USA 97, 10044–10049. Gaffield, W., and Yamada, G. (2001). Unique functions of Sonic hedge- 56. Shbailat, S.J., Khila, A., and Abouheif, E. (2010). Correlations between hog signaling during external genitalia development. Development spatiotemporal changes in gene expression and apoptosis underlie 128, 4241–4250. wing polyphenism in the ant Pheidole morrisi. Evol. Dev. 12, 580–591. 35. Yamada, G., Satoh, Y., Baskin, L.S., and Cunha, G.R. (2003). Cellular and 57. Harris, M.P., Fallon, J.F., and Prum, R.O. (2002). Shh-Bmp2 signaling molecular mechanisms of development of the external genitalia. module and the evolutionary origin and diversification of feathers. Differentiation 71, 445–460. J. Exp. Zool. 294, 160–176. Current Biology Vol 23 No 12 1074

58. Abzhanov, A., Protas, M., Grant, B.R., Grant, P.R., and Tabin, C.J. (2004). Bmp4 and morphological variation of beaks in Darwin’s finches. Science 305, 1462–1465. 59. Schneider, R.A., and Helms, J.A. (2003). The cellular and molecular origins of beak morphology. Science 299, 565–568. 60. Itoh, Y., Suzuki, M., Ogawa, A., Munechika, I., Murata, K., and Mizuno, S. (2001). Identification of the sex of a wide range of Carinatae birds by PCR using primer sets selected from chicken EE0.6 and its related sequences. J. Hered. 92, 315–321. 61. Nieto, M.A., Patel, K., and Wilkinson, D.G. (1996). In situ hybridization analysis of chick embryos in whole mount and tissue sections. Methods Cell Biol. 51, 219–235. 62. Laufer, E., Pizette, S., Zou, H., Orozco, O.E., and Niswander, L. (1997). BMP expression in duck interdigital webbing: a reanalysis. Science 278, 305.