Are current critiques of the theropod origin of birds science? Rebuttal to Feduccia (2002)

Auk, The, Apr 2003 by Prum, Richard O

Feduccia stated that the scale-to-feather model "conforms nicely to what we know about feather embryology." That is patently incorrect. The dorsal and ventral surfaces of a scale form from the dorsal and ventral surfaces of the first primordial scale outgrowth, whereas the obverse (outer) and reverse (under) surfaces of a pennaceous feather develop from the outer and inner surfaces of the tubular feather germ, respectively (Fig. 3) (Davies 1889, Lucas and Stettenheim 1972, Prum 1999). As Davies (1889) first pointed out, because of the topology of feather development the two surfaces of a planar feather are not homologous with the two surfaces of a planar scale, and could not have evolved from them (Prum 1999, Prum and Brush 2002). Unfortunately, that fundamental fact was ignored by Heilmann (1926), who revived the elongate scale theory of feather evolution, and by the generations of ornithologists who subsequently advocated elongate scale hypotheses (Lowe 1935, Becker 1959, Parkes 1966, Maderson 1972, Regal 1975, Dyck 1985, Martin and Czerkas 2000), including two decades of publications by Feduccia (1980, 1985, 1993, 1996, 1999a, b). Elongate scale theories of feather evolution were falsified on the basis of developmental evidence more than a century ago, and no proponent of the elongate scale theory has ever countered Davies' (1889) fatal developmental observations (Prum 1999, Prum and Brush 2002). Furthermore, no elongate scale model has ever presented a satisfactorily detailed hypothesis of the evolution of feather branched structure (Prum and Brush 2002). Any theory of the origin of feathers should also explain the evolution of how feathers grow, and elongate scale theories have entirely failed that test.

Elsewhere (Prum 2001, Prum and Brush 2002), I have criticized the recent hypothesis that the elongate scales of Longisquama are homologous with feathers (Jones et al. 2000). Feduccia (2002) criticized my observations of Longisquam as cursory, but he neglected to mention that he has repeatedly published observations that are entirely in agreement with mine (Prum 2001). For example, he wrote, "In 1982 I examined the specimen of Longisquama in Moscow and could see no indication that the elongated scales were particularly feather-like" (Feduccia 1985:76). In 1999, he wrote, "No doubt... the scales of Longisquama were not transmuted into feathers, but the specimen does show the tremendous experimentation in feather-like scales in the basal archosaurs before the advent of feathers" (Feduccia 1999b: 133). Somehow, he made a complete and rapid conversion from thinking that Longisquama was a "bizarre and unique solution to the problem of gliding" (Feduccia 1999b: 95) to thinking that Longisquama is the closest known relative of birds (Jones et al. 2000).

Although Feduccia complimented our recent experimental research on feather development, he failed to grasp the consequences of our data (Harris et al. 2002). We demonstrated that feathers evolved through a series of cooptions-evolutionary reutilizations-of a plesiomorphic molecular developmental molecular module (Fig. 4). The identified elements of that molecular module consists of a pair of intercellular signaling genes Sonic Hedgehog (Shh) and Bone Morphogenetic Protein 2 (Bmp2). Shh and Bmp2 have conserved set of developmental interactions within the integument, like mechanical components of a machine. This molecular module has been repeatedly recruited for controlled morphogenesis of novel structures during the evolution of feathers and feather complexity (Fig. 4). We have shown that in the first stage, feather placodes share an anterior-posterior polarized expression of Shh and Bmp2 with the placodes of bird scales and alligator scales, but that all subsequent stages of feather development are derived and unique to feathers (Fig. 4). Then the feather bud develops by distal co-expression of Shh and Bmp2, creating the tubular feather germ (Fig. 4, Event 1). Subsequently, barb ridges develop through longitudinal stripes in Shh and Bmp2 expression (Fig. 4, Event 2). Ultimately, the ventral new barb locus and the rachis develop through the coordinated ventral bifurcation and dorsal cessation of Shh and Bmp2 signaling stripes (Fig. 4, Event 3).