Systematic revision of the avian family Cisticolidae based on a multi-locus phylogeny of all genera

Complete species-level phylogeny of the leaf warbler (Aves: Phylloscopidae) radiation

The leaf warbler radiation (Aves: Phylloscopidae) has undergone a c. 50% increase in the number of recognised species over the last three decades, mainly as a result of analyses of vocalisations and DNA. Using a multilocus dataset for all of the species in this family, and multispecies coalescent-based as well as concatenation methods, we provide the first complete species-level phylogeny for this important group, as well as an estimate of the timing of diversification. The most recent common ancestor for the family was dated at 11.7 million years ago (mya) (95% highest posterior density 9.8-13.7 mya), and divergence times between sister species ranged from 0.5 mya (0.3-0.8 mya) to 6.1 mya (4.8-7.5 mya). Based on our results, we support synonymising Seicercus with Phylloscopus, which results in a monogeneric Phylloscopidae. We discuss the pros and cons of this treatment,and we argue againstproliferation of taxonomic names,and conclude that a large monogeneric Phylloscopidae leads to the fewest taxonomic changes compared to traditional classifications. We briefly discuss morphological evolution in the light of the phylogeny. The time calibrated phylogeny is a major improvement compared to previous studies based on a smaller number of species and loci and can provide a basis for future studies of other aspects of phylloscopid evolution.

The systematics and biogeography of African tailorbirds (Cisticolidae: Artisornis) with comment on the choice of Bayesian branch-length prior when analyzing heterogeneous data

The Long-billed Tailorbird (Artisornis moreaui), one of Africa's rarest birds, has a strikingly disjunct distribution, the origin of which has long puzzled biogeographers. One small population (subspecies moreaui) occurs in sub-montane forest in the East Usambara Mountains, a sky island near the coast of northern Tanzania, and another (subspecies sousae) on Serra Jeci in northwestern Mozambique, 950km away. The African Tailorbird, the putative sister-species of Long-billed Tailorbird, also occurs in the East Usambara Mountains and on Serra Jeci, but in addition occupies all the Eastern Arc Mountain forests between these disjunct sites. Stuart (1981) hypothesized that the two tailorbird distributions could be explained by strong ecological competition, with African Tailorbird populations having eliminated Long-billed Tailorbird populations via competitive exclusion in montane forests between the East Usambara and Serra Jeci. If such competitive exclusion explains these geographic distributions, the co-occurrence of the two species in the East Usambara and at Serra Jeci may be ephemeral, with the status of Long-billed Tailorbird especially in doubt. We sought to (1) determine whether the two species of African tailorbirds are indeed sister-species, and (2) test predictions from Stuart's (1981) competitive exclusion hypothesis using genetic data. Phylogenetic analyses of our seven gene dataset (3 mtDNA, 4 introns; 4784bp) indeed place these two species together in the genus Artisornis. Instead of finding shallow divergence among African Tailorbird populations and deep divergence between Long-billed Tailorbird populations as expected from Stuart's hypothesis, we recover deep genetic divergence and geographic structure among populations of both tailorbird species. This result is consistent with long-term co-existence of the two species at East Usambara and Serra Jeci. Observational data from both the East Usambara and Serra Jeci suggest that the two species have diverged in use of forest canopy strata. From a conservation standpoint, our results suggest that extinction of the Long-billed Tailorbird as a function of competition with African Tailorbird is highly unlikely, and should not be viewed as imminent. Threats to its survival are instead anthropogenic, and conservation measures should take this into account. Finally, our empirical results suggest that mis-specification of the branch-length prior in Bayesian analyses of mitochondrial DNA data can have a profound effect on the overall tree-length (sum of branch-lengths), whereas the topology and support values tend to remain more stable. In contrast, mis-specification of the branch-length prior had a lesser impact on all aspects of the nuclear-only DNA analyses. This problem may be exacerbated when mitochondrial and nuclear DNA analyses are combined in a total evidence approach.