Norell and Clarke reply

Feduccia comments that other Mesozoic bird specimens are more, or just as, useful for tackling questions concerning the origin of extant bird lineages. Although all specimens contain some information, we disagree with Feduccia's current assertion that Apsaravis is simply one of a group of “abundant” ornithurine fossils. The specimens he mentions are either not ornithurine or are so poorly preserved that they have not shed much light on their own phylogenetic positions, let alone on broader patterns of avian evolution.

The two “ornithurine” birds Liaoningornis and Chaoyangia fall outside Ornithurae in Feduccia's own work. The explicit cladistic definition of Ornithurae (most recent common ancestor of Hesperornithiformes plus Aves and all descendants2) is less inclusive than Feduccia's more subjective definition (taxa other than those that are not 'modern' enough to be ornithurine). Feduccia's 'Ornithurae' is predicated on the existence of a 'Sauriurae' or the paraphyletic group that contains these primitive taxa. The fact that Apsaravis and our analyses add to the mounting evidence3,4 against sauriurine monophyly has been overlooked in Feduccia's estimation of the importance of Apsaravis.

Other specimens that we did not consider are problematic and underscore the importance of well preserved and phylogenetically placed taxa such as Apsaravis. Feduccia did not include the fragmentary Otogornis, Ambiortus and Gansus in his analyses5. Although he claims that Chaoyangia possesses a “toothed skull”, the holotype actually consists only of a torso and partial hindlimbs. The “toothed skull” belongs to a specimen once referred to as Chaoyangia5 but later identified as the holotype of Songlingornis linghensis6. This specimen cannot be referred to Chaoyangia (as indeed it has not been6) as no element known from the holotype is also represented in the referred specimen.

Although we did not comment on the implications of Apsaravis for the timing of the origin of Aves, Feduccia's conjecture that it cannot inform our understanding of this origin (because it is not part of an extant lineage) is incompatible with his own arguments. In recounting the origin of Aves, he invokes taxa such as ichthyornithiforms and hesperornithiforms, which are not parts of extant lineages. Furthermore, it has been argued7 that gap analyses may be consistent with Cretaceous or Tertiary diversification of avian lineages, depending on what model of diversification rate and recovery potential is considered realistic.

Reasoning derived from phylogenetic analysis is a powerful way to test hypotheses of relationships or the evolution of morphology (for example, enantiornithine monophyly and novelties in the flight apparatus). We used a phylogenetic test to assess the idea that transitional 'shore birds' gave rise to all extant birds through an ecological bottleneck1.

If such a bottleneck occurred, then when ecology is bracketed phylogenetically for living birds, 'shore bird' morphology and ecology should be basal to the crown clade, as well as in its nearest sister taxa. However, virtually all molecular and morphological evidence places 'land birds' (tinamous, ratites, galliforms and anseriforms, for example) at the base of Aves8,9. Charadriiformes, the extant lineage referred to as shore birds, are placed as derived forms within Aves8,9. Thus, if the ecologies that are basal to the crown clade are bracketed, no support is found for such a bottleneck.

Apsaravis, because of its phylogenetic placement, constrains the inference of the ecologies of the most recent common ancestor of the avian crown clade. We do not understand how an ornithurine with no 'shore bird' morphologies, from a dunefield10, can be interpreted as compatible with Feduccia's idea1 of ecological restriction of these taxa to shorelines and marine environments. If Apsaravis can simply be assumed, without consideration of phylogenetic tests, to have flown from an unknown nearby lake, then we do not see how Feduccia's hypothesis is testable.