RHINOCRYPTIDAE

Maurício, Giovanni Nachtigall, Areta, Juan Ignacio, Bornschein, Marcos Ricardo & Reis, Roberto E., 2012, Morphology-based phylogenetic analysis and classification of the family Rhinocryptidae (Aves: Passeriformes), Zoological Journal of the Linnean Society 166 (2), pp. 377-432 : 419-420

RELATIONSHIPS WITHIN RHINOCRYPTIDAE

Several previous hypotheses of relationships within Rhinocryptidae were corroborated in the present study, but others were not. The polytypic genera Merulaxis , Eleoscytalopus , Pteroptochos , and Scytalopus were recovered as monophyletic in the analysis, but because only one of the two species of Scelorchilus was included here, its monophyly remains to be properly tested with the inclusion of S. albicollis .

The paraphyletic nature of Scytalopus with respect to the inclusion of the white-bellied taxa ‘S’ indigoticus and ‘S’ psychopompus in the genus, as shown by Maurício et al. (2008), was corroborated here, thus supporting the erection of the genus Eleoscytalopus to place these two species. This proposition was based primarily upon a molecular phylogenetic analysis that recovered an Eleoscytalopus + Merulaxis clade, but also included some syringeal and osteological data from an early phase of the project that resulted in the present study (see Maurício et al., 2008). The syringeal character-states suggested by these authors as support for a sister-taxon relationship between Eleoscytalopus and Merulaxis – cranial portion of the Processus vocalis with a soft consistency and A3–A5 elements dorsally reduced/absent – were optimized here together with six others (four syringeal and two osteological; see above) as synapomorphies for a clade containing exclusively these two genera, although the placement of this clade within the Rhinocryptidae was considerably distinct between the two studies ( Fig. 43 ).

A close relationship between Scelorchilus and Pteroptochos has been suggested by early taxonomists (e.g. Sclater, 1874) and was recently corroborated by molecular data ( Chesser, 1999; Maurício et al., 2008; Moyle et al., 2009; Ericson et al., 2010). In the present study this relationship was strongly supported, with seven synapomorphies, none of which has been previously mentioned as evidence of a relationship between these genera. One of these synapomorphies, presence of a dorsoventral intrinsic muscle in the syrinx, was described as having a generalized occurrence in the Rhinocryptidae ( Ames, 1971) , and was suggested as a synapomorphy for the family as a whole ( Rice, 2005). This hypothesis is herein refuted as this dorsally originating intrinsic muscle actually has a very restricted occurrence in the family (i.e. only in Acropternis , Scelorchilus , and Pteroptochos ).

A Rhinocrypta + Teledromas clade was recovered by Moyle et al. (2009) and the present study, but Ericson et al. (2010) found Teledromas as sister to a Rhinocrypta + Acropternis clade. Morphological evidence in favour of the Rhinocrypta + Teledromas arrangement was solid. Besides being defined by seven synapomorphies, a general similarity in size and proportions of the sternum, cranium, and pelvis characterizes the members of the Rhinocrypta + Teledromas clade. Two of the seven synapomorphic conditions supporting this branch were previously described for Rhinocrypta only, namely parietal and frontal bones fully pneumatized ( Feduccia & Olson, 1982; Krabbe & Schulenberg, 2003) and postorbital and zygomatic processes fused ( Claramunt & Rinderknecht, 2005).

On the basis of external similarities it has been suggested that the genera Merulaxis , Eugralla , Myornis , and Scytalopus form a clade, and that within this clade Merulaxis and Myornis would be sister-taxa ( Krabbe & Schulenberg, 1997, 2003; Irestedt et al., 2002). Recent molecular studies recovered such a clade, but Merulaxis was sister to the recently described genus Eleoscytalopus instead of to Myornis ( Maurício et al., 2008; Mata et al., 2009; Ericson et al., 2010). The present study partially corroborated the molecular findings, as it recovered Eleoscytalopus as sister to Merulaxis and also a well-supported Eugralla + Myornis + Scytalopus clade, but these two clades were placed in very distinct points of the rhinocryptid morphology-based tree: whereas the former is a basal branch the latter is an apical clade embeded within a clade containing the large-bodied genera. Of the six synapomorphies that define the Eugralla + Myornis + Scytalopus clade only the unfused clavicles were previously mentioned as a supporting character for this grouping ( Maurício et al., 2008; see also Feduccia & Olson, 1982). Within this clade all molecular phylogenies recovered Eugralla as sister to Scytalopus and Myornis as basal to both ( Maurício et al., 2008; Mata et al., 2009; Moyle et al., 2009; Ericson et al., 2010), thus differing from the morphological data which showed Eugralla as sister to a Myornis + Scytalopus clade. The morphological data did not allow us to test previous hypotheses of relationships in the genus Scytalopus (e.g. Arctander & Fjeldså, 1994; Krabbe & Schulenberg, 1997; Bornschein et al., 1998; Maurício, 2005; Mata et al., 2009). The strict consensus of most parsimonious trees could not even recover the phylogenetic subdivision of the genus into an Andean and a Brazilian component as proposed by Mata et al. (2009).

The present study did not corroborate the division of the family into the subfamilies Rhinocryptinae and Scytalopodinae as proposed by Moyle et al. (2009). The comprehensive molecular phylogeny of Ericson et al. (2010) also diverged from the results of the former authors in that the Scytalopodinae, to be a monophyletic group, should include Scelorchilus and Pteroptochos , both being part of Rhinocryptinae sensu Moyle et al. (2009) , as well as Eleoscytalopus and Merulaxis , not sampled in the latter study. The genera Liosceles and Psilorhamphus formed a clade in Ericson et al. (2010) that was sister to a Teledromas + Rhinocrypta + Acropternis clade, thus approaching the composition of the Rhinocryptinae as recovered by Moyle et al. (2009) who, however, did not sample Psilorhamphus . In contrast to these molecular phylogenies, the basal relationships recovered by the present morphological analysis consisted of Liosceles and Psilorhamphus as being successively basal to a clade containing the remaining ten genera, within which Eleoscytalopus + Merulaxis were sister to a group composed of the remaining eight genera. Of this branching scheme, the basal position of Liosceles relative to the rest of the family and the grouping of eight genera were supported by four or six synapomorphies and relatively strong Bremer values (i.e. 4), and thus at least these nodes may be regarded as good topological alternatives of the deeper rhinocryptid cladogenesis relative to the molecular findings.

In summary, clades supported by six or more synapomorphies and Bremer values of 6 or 7, such as Eleoscytalopus + Merulaxis , Scelorchilus + Pteroptochos , Rhinocrypta + Teledromas , and Eugralla + Myornis + Scytalopus ( Fig. 43 ), were the main points of congruence between the present morphological phylogeny and the previous phylogenetic work with the family, all sequence-based. On the other hand, more inclusive nodes (i.e. those including more than three genera) were dissimilar between this study and the molecular phylogenies, although two of those nodes received Bremer support values of 4 in the morphological analysis. None of the synapomorphies supporting these basal nodes was previously mentioned in the literature.

THE IMPORTANCE OF MORPHOLOGY IN PHYLOGENETIC INFERENCE

The differences between the results of Moyle et al. (2009) and Ericson et al. (2010) regarding the placement of some genera, in addition to the substantial topological differences found by the latter authors when the three nuclear genetic markers are analysed separately ( Ericson et al., 2010: 343, fig. 3), revives the debate concerning species trees versus gene trees: ‘If the evolution of a gene differs from that of a species, trees reconstructed from molecular data may give well-supported wrong answers to questions about species phylogeny’ ( Hillis & Wiens, 2000) and ‘There are many factors that may cause molecular analyses to reconstruct clades that are both incorrect and statistically well supported...’ ( Wiens, 2004). In this context, it is important to bear in mind that ‘a typical set of morphological characters should draw on information from many different unlinked genes [...], whereas the characters in a given molecular data set are often linked and inherited as a single unit’ ( Wiens, 2004). Therefore, as Wiens (2004: 654) states, given that we are not at a stage where all molecular phylogenies can be reconstructed without error, it is important to have rigorous morphology-based phylogenies as a ‘reality check’ for molecular results. It is also important to consider the more complex (i.e. less parsimonious) evolutionary pathway of morphological characters implied by the molecular phylogenies. For example, the phylogenetic placement of Liosceles both in Moyle et al. (2009) and in Ericson et al. (2010) implies reversals or parallel transformations in several characters, among which are complex ones such as the presence/absence of an osseous wall in the fossa pneumotricipitalis (character 47) and fusion of the dorsal iliac crests to form the Crista iliosynsacralis (character 49). Likewise, although the clade Teledromas + Rhinocrypta was supported by seven morphological synapomorphies, had a Bremer support of 7 and was recovered by Moyle et al. (2009), it was never recovered by Ericson et al. (2010). The clade ( Teledromas ( Acropternis + Rhinocrypta )) recovered by the combined analysis of Ericson et al. (2010) would imply reversals in Acropternis or independent evolution in Teledromas and Rhinocrypta of seven character states, including four that are exclusive to the clade formed by these two genera in the morphological tree. Here and again, we echo the claims of Wiens (2004: 654) about the importance of having rigorous morphology-based phylogenies as a ‘reality check’ for molecular results.

Figures

Figure 43. Strict consensus of 7428 equally most parsimonious trees derived from the cladistic analysis of the character state matrix in Appendix 2. On each branch, solid circles indicate exclusive synapomorphies and open circles homoplastic traits; numbers above circles represent characters as numbered in the character analysis section, with correspondig character-states appearing below circles. All characters are non-additive (except char. 74, whose states are ordered based on ontogenetic evidence) and unweighted. Bremer support indices are shown in parentheses above character numbers. Vertical bars on the right indicate family boundaries: A, Tyrannidae; B, Conopophagidae; C, Melanopareiidae; D, Grallariidae; E, Thamnophilidae; F, Formicariidae; G, Scleruridae; H, Furnariidae; I, Dendrocolaptidae; J, Rhinocryptidae.