Biogeography and Diversification of Rhegmatorhina (Aves: Thamnophilidae): Implications for the Evolution of Amazonian Landscapes During the Quaternary

DOI: 10.1111/jbi.13169

ORIGINAL ARTICLE

Biogeography and diversification of Rhegmatorhina (Aves: Thamnophilidae): Implications for the evolution of Amazonian landscapes during the Quaternary

1Coord. de Biodiversidade, Instituto Nacional de Pesquisas da Amazonia,^ Abstract Manaus, AM, Brazil Aim: To test the importance of alternative diversification drivers and biogeographi- 2Department of Ornithology, American cal processes for the evolution of Amazonian upland forest birds through a densely Museum of Natural History, New York, NY, USA sampled analysis of diversification of the endemic Amazonian genus Rhegmatorhina 3Coord. de Zoologia, Museu Paraense at multiple taxonomic and temporal scales. Emılio Goeldi, Belem, PA, Brazil Location: Amazonia. 4Graduate Program in Genetics, Conservation and Evolutionary Biology, Taxon: Antbirds (Thamnophilidae). INPA, Manaus, AM, Brazil Methods: We sequenced four mtDNA and nuclear gene regions of 120 individuals 5Museum of Natural Science, Louisiana State University, Baton Rouge, LA, USA from 50 localities representing all recognized species and subspecies of the genus. We performed molecular phylogenetic analyses using both gene tree and species Correspondence Camila C. Ribas, Coord. de Biodiversidade, tree methods, molecular dating analysis and estimated population demographic his- Instituto Nacional de Pesquisas da tory and gene flow. Amazonia,^ Manaus, AM, Brazil. Email: [email protected] Results: Dense sampling throughout the distribution of Rhegmatorhina revealed that the main Amazonian rivers delimit the geographic distribution of taxa as inferred Present addresses from mtDNA lineages. Molecular phylogenetic analyses resulted in a strongly sup- Chrysoula Gubili, Hellenic Agricultural Organization, Fisheries Research Institute, ported phylogenetic hypothesis for the genus, with two main clades currently sepa- Nea Peramos, Kavala, Macedonia, Greece. rated by the Madeira River. Molecular dating analysis indicated diversification

Funding information during the Quaternary. Reconstruction of recent demographic history of populations Fundacßao~ de Amparo a Pesquisa do Estado revealed a trend for population expansion in eastern Amazonia and stability in the do Amazonas, Grant/Award Number: 929/ 2011; Fundacß~ao de Amparo a Pesquisa do west. Estimates of gene flow corroborate the possibility that migration after diver- Estado de Sao~ Paulo, Grant/Award Number: gence had some influence on the current patterns of diversity. 2007/07637-3, 2012/50260-6; Conselho Nacional de Desenvolvimento Cientıfico e Main Conclusions: Based on broad-scale sampling, a clarification of taxonomic Tecnologico, Grant/Award Number: boundaries, and strongly supported phylogenetic relationships, we confirm that, first, 308927/2016-8, 310880/2012-2, 470610/ 2008-5; United States Agency for mitochondrial lineages within this upland forest Amazonian bird genus agree with International Development, Grant/Award spatial patterns known for decades based on phenotypes, and second, that most lin- Number: PEER Cycle 5,AID-OAA-A-11- 00012; Coordenacß~ao de Aperfeicßoamento de eages are geographically delimited by the large Amazonian rivers. The association Pessoal de Nıvel Superior, Grant/Award between past demographic changes related to palaeoclimatic cycles and the histori- Number: 23038.001963/2012-41; Directorate for Biological Sciences, Grant/ cally varying strength and size of rivers as barriers to dispersal may be the path to Award Number: 1146423, 1241066; F.M. the answer to the long-standing question of identifying the main drivers of Amazo- Chapman Fund/AMNH; NSF, Grant/Award Number: 1241066, 1146423; NASA nian diversification.

Editor: Miles Silman

KEYWORDS Amazonia, Antbirds, diversification, molecular systematics, Neotropic, phylogeny, phylogeography, taxonomy

1 | INTRODUCTION Recent, more detailed phylogenetic and phylogeographic studies, associated systematic revisions and temporal diversification analyses Amazonian diversity and its distribution are still poorly known, even of extant Amazonian taxa show constant diversification rates in well-studied groups such as birds (Whitney & Cohn-Haft, 2013). throughout the Neogene/Quaternary (Brumfield, 2012; d’Horta, Numerous regions of this vast forest lack modern biotic inventories, Cuervo, Ribas, Brumfield, & Miyaki, 2013; Patel, Weckstein, Patane, and many taxa remain to be described (Barrowclough, Cracraft, Bates, & Aleixo, 2011). Detailed taxonomic sampling and phyloge- Klicka, & Zink, 2016; Hopkins, 2007). Despite gaps in our taxonomic netic analysis have also shown a Plio–Pleistocene origin of species of and distributional knowledge, Amazonia harbours the highest species the genus Psophia, whose ranges are clearly delimited today by the diversity of any biome on Earth (de Groot et al., 2012). However, large Amazonian rivers (Ribas et al., 2012), indicating that the our understanding of its evolutionary origins remains unclear dynamic history of the Amazonian drainage system may be an because an answer relies on fundamental knowledge about the important driver of diversification during this period (Nogueira et al., region’s constituent taxa, their distributions and their interrelation- 2013). Population-level sampling of Psophia species also shows dis- ships, for which there are still many gaps. Multiple studies have tinct demographic signatures for populations from different Amazo- shown that currently recognized species significantly underestimate nian areas of endemism, suggesting differential effects of climatic the number of evolutionary taxa that are relevant for analyses of cycles on forest bird populations (Cheng et al., 2013; Ribas et al., diversification and biogeography (e.g. Fernandes, Wink, & Aleixo, 2012; Wang et al., 2017). Additional studies of other groups of Ama- 2012; Harvey & Brumfield, 2015; Ribas, Aleixo, Nogueira, Miyaki, & zonian birds corroborate these findings, supporting both drainage Cracraft, 2012; Ribas, Joseph, & Miyaki, 2006; Thom & Aleixo, evolution and climatic-driven changes on vegetation cover as impor- 2015). Thus, how we think about species-level taxa and how well tant drivers of diversification during the Plio–Pleistocene (d’Horta each group is geographically sampled and taxonomically known is et al., 2013; Fernandes, Wink, Sardelli, & Aleixo, 2014; Fernandes central to solving the question of the evolutionary drivers of Amazo- et al., 2012; Ferreira, Aleixo, Ribas, & Santos, 2017; Schultz et al., nian diversity. 2017; Thom & Aleixo, 2015). However, incongruent patterns (Smith An important uncertainty is how the origin of this high diversity et al., 2014) suggest a complex history and a combination of biogeo- relates to landscape history, which has been the subject of consider- graphic processes (including dispersal and vicariance) acting across able debate over many decades (Haffer, 1969; Hoorn et al., 2010; multiple time scales. Ribas et al., 2012; Smith et al., 2014). Our growing knowledge of The genus Rhegmatorhina includes taxa specialized for army-ant the geological and palaeoclimatological history of Amazonia (Cheng following in the understory of tall upland Terra Firme Amazonian for- et al., 2013; Hoorn et al., 2010; Latrubesse et al., 2010; Nogueira, est (Willis, 1969). There are five recognized species—R. gymnops, Silveira, & Guimar~aes, 2013; Wang et al., 2017) is now allowing us R. hoffmannsi, R. berlepschi, R. melanosticta and R. cristata (Zimmer & to ask specific questions about patterns of biotic history (Baker Isler, 2003). Large rivers define the majority of ranges for these spe- et al., 2014). In particular, how past climate variation may have cies, which occur in five Amazonian areas of endemism (sensu affected the demography and connectivity of forest-adapted popula- Borges & Silva, 2011; Cracraft, 1985; Silva et al. 2005): Tapajos, tions, and how landscape reorganization relating to drainage evolu- Rondonia,^ Inambari, Napo and Jau. The genus is absent from the tion may have contributed to the origin of upland forest species. three easternmost areas: Guiana, Xingu and Belem. Two species Traditionally, these candidate palaeoenvironmental drivers are sup- (R. hoffmannsi and R. berlepschi) occupy the Rondonia^ area, and one posed to have operated at different time frames: glacial cycles species (R. melanosticta) occupies two areas (Inambari and Napo). occurred mostly during the Quaternary (Haffer, 1969), whereas the Although species distributions within the genus are delimited by current configuration of the Amazonian drainage system has been major Amazonian rivers, hybridization has been documented largely assumed to date to the Miocene (Hoorn et al., 2010), despite between R. gymnops and R. hoffmannsi in the headwaters of the evidence pointing to more recent times (Latrubesse et al., 2010; Tapajos River, which is the limit between the Tapajos and Rondonia^ Nogueira et al., 2013). Due to this temporal difference, the two dri- areas (Weir, Faccio, Pulido-Santacruz, Barrera-Guzman, & Aleixo, vers are rarely integrated. Therefore, studies have assumed that spe- 2015). Here, we present combined phylogeographic and phyloge- cies may be old, with their origin related to Miocene drainage netic multilocus analyses for the genus to test the importance of evolution (Hoorn et al., 2010), or young and their origin related to alternative diversification drivers and biogeographical processes that Quaternary glacial cycles, a very simplistic approach as pointed out have been suggested by previous studies on Amazonian upland by Rull (2015). forest birds. With extensive sampling both in terms of range RIBAS ET AL. | 919 representation and number of individuals, we address intraspecific 2.2 | DNA extraction, amplification and data genetic structure, refine taxonomic and lineage distribution limits, analysis and estimate the demographic history of populations. We integrate genetic analyses at both intra- and interpopulational scales to under- Genomic DNA was extracted using the DNeasy Blood and Tissue kit stand spatially and temporally the history of this genus, comparing (QIAGEN). Three mitochondrial genes (cytochrome b, cytb; NADH its pattern with other Amazonian groups, and relate these results to dehydrogenase subunit 2, ND2; NADH dehydrogenase subunit3, the different hypotheses about Amazonian landscape evolution. ND3) and one nuclear intron (b-fibrinogen intron 5, FIB5) were sequenced (for the primers used see Table S2.1, Appendix S2). Sequences were aligned, gametic phase was estimated, recombina- 2 | MATERIALS AND METHODS tion tests were performed, and summary statistics were calculated (see Appendix S2 for details). Statistically significant differences in 2.1 | Sampling genetic diversity indices (haplotype and nucleotide diversities) among We sampled 120 specimens representing a broad geographic cover- species for each marker were tested using a two-way ANOVA test age of the distribution of all seven named taxa within Rhegmatorhina: in R 3.0.3 (R Core Team 2014). the three subspecies of R. melanosticta plus the remaining four monotypic species (Figure 1, Table S1, Appendix S1 in Supporting 2.3 | Phylogenetic analyses and haplotype Information). We also included sequences of the five Gymnopithys networks species (G. salvini, G. bicolor, G. leucaspis, G. lunulatus and G. rufigula) as this genus has been shown to be paraphyletic, with G. leucaspis All analyses were performed using a concatenated mtDNA matrix and G. rufigula forming a sister clade to Rhegmatorhina (Isler, Bravo, with partitions and models selected in PARTITIONFINDER 1.1.1 (Lanfear, & Brumfield, 2014). Phlegopsis nigromaculata was included as the Calcott, Ho, & Guindon, 2012). Bayesian phylogenetic inference (BI) out-group. was performed using MRBAYES 3.2.2 (Ronquist et al., 2012), with two

FIGURE 1 Sampling localities and molecular phylogeny of the genus Rhegmatorhina based on Bayesian analysis of the mtDNA genes Cyt b, ND2 and ND3 (2,440 bp). Maximum likelihood analysis yields the same topology. Numbers on nodes are posterior probabilities/ML bootstrap. *indicates maximum support values [Colour figure can be viewed at wileyonlinelibrary.com] 920 | RIBAS ET AL. independent runs of 5 9 106 generations and four Markov chain 3.2 | Phylogenetic inferences Monte Carlo (MCMC) chains. Convergence was assessed in TRACER 1.6 (Rambaut, Suchard, Xie, & Drummond, 2014). Maximum likeli- The mitochondrial data set strongly supported the monophyly of hood analyses (ML) were performed in PHYML 3.0 (Guindon et al., Rhegmatorhina, and a sister relationship to the clade formed by (G. 2010), with 1,000 bootstrap replications. Genealogical relationships bicolor,(G. leucaspis, G. rufigula)) (Figure 1). Within Rhegmatorhina, among haplotypes were estimated using median-joining networks there are two well-supported clades separated by the Madeira River (Bandelt, Forster, & Rohl,€ 1999). (Figure 1). The western clade includes R. cristata, from the Jau area of endemism (Japura-Negro interfluve, Borges & Silva, 2011) and R. melanosticta, from western Amazonia, which was subdivided into two 2.4 | Species tree and divergence time estimates main clades separated by the Solimoes~ and Ucayali Rivers (Figure 1). A species tree was reconstructed using the multilocus data set in The eastern clade includes R. gymnops from the Tapajos area of ende-

*BEAST, as implemented in BEAST 1.8.1 (Heled & Drummond, 2010). mism (Tapajos-Xing u interfluve), and R. hoffmannsi and R. berlepschi The seven main mitochondrial lineages (Figure 1) were assigned as from the Rondonia^ area of endemism (Madeira-Tapajos interfluve). ‘species’ in the analysis. The lognormal distribution for the relaxed The recovered sister relationship between R. hoffmannsi and R. berlep- uncorrelated rates and Yule process were used for all genes. The schi had moderately high support (0.98/72). Two strongly supported substitution rate of cytochrome b (2.1% divergence per million years, clades were also found within R. hoffmannsi, yet they currently are not Weir & Schluter, 2008) was used for time calibration, and rates for separated by any geographic barrier and co-occur in locality H5 the remaining markers were estimated. We ran the analysis for (Cachoeira de Nazare, on the West bank of Rio Ji-Parana, Figure 1). 2 9 108 generations and sampled every 1,000 generations. Another

*BEAST analysis was run including the five Gymnopithys species (see 3.3 | Haplotype networks, coalescence and Figure 1) to date the origin of the Rhegmatorhina lineage. phylogenetic information As there was extensive Fib5 allele sharing among species (see

Results), and the *BEAST species tree method does not account for No shared haplotypes among the seven lineages were found in the gene flow after isolation (Heled & Drummond, 2010), divergence mtDNA markers (Figure 2a–c). In contrast, the nuclear gene revealed times were also estimated in a gene tree framework, using only the one most common haplotype found in all species but R. cristata, plus mtDNA genes in BEAST 1.8.1 (Drummond & Rambaut, 2007). The three shared haplotypes between the two R. hoffmannsi lineages, analysis was run for 5 9 107 generations and time calibrated as one between R. melanosticta and R. cristata, and one between R. described above. hoffmannsi and R. berlepschi (Figure 2d). In addition, divergence times and migration rates were co-esti- These results, and our dating analysis discussed below, indicate that mated under an isolation-with-migration model (Hey & Nielsen, diversification of crown Rhegmatorhina is relatively young, dating to 2004), implemented by IMa2 (Hey, 2010). We performed the analy- less than ~2.0 million years ago (Ma). The haplotype networks show sis independently for pairs of taxa with adjacent distributions (total that taxonomic structure exists in mitochondrial DNA but not in the of eight pairs, see Appendix S2 for details). nuclear marker, consistent with population genetic predictions of their relative coalescent times (Hung, Drovetski, & Zink, 2016; Zink & Bar- rowclough, 2008). Our results for Rhegmatorhina are consistent with 2.5 | Demographic analysis those found in the Amazonian clade Psophia (Ribas et al., 2012) in Significant deviations from neutral evolution and constant population which a nuclear marker was only capable of resolving the oldest diver- sizes were tested for using summary statistics (see Appendix S2 for gence (~2.0–2.7 Ma). Species of Rhegmatorhina are common monoga- details). Historical fluctuations of population sizes were estimated mous understory birds and would be expected to have a larger using extended Bayesian skyline plots (EBSPs) in BEAST (Heled & effective population size (Ne) relative to species of Psophia, which have Drummond, 2010) with fifty million generations and an uncorrelated smaller census populations and are co-operatively polyandrous, thus lognormal molecular clock with the cytb substitution rate of 2.1% further suggesting that nuclear loci would not be predicted to coalesce per million years (Weir & Schluter, 2008). in Rhegmatorhina over this short time-period. Thus, for reasons of age and population size, mitochondrial DNA is more appropriate than a single or a few nuclear markers for recovering evolutionary relationships 3 | RESULTS within the Rhegmatorhina species-complex. A genomic perspective on the evolution of the genus will only be informative if it relies on thou- 3.1 | Sequence data and genetic diversity sands of independently segregating loci (Harvey and Brumfield 2015). A total of 2,945 bp were sequenced (Table S2.2 in Appendix S2). The nuclear locus showed no recombination (p = .684) and higher 3.4 | Species tree and divergence time estimates intraspecific nucleotide diversity (Table S2.3 in Appendix S2). Rheg- matorhina melanosticta A and B had significantly higher genetic diver- The species tree corroborated the presence of two well-supported sity for cytb and ND2 (p-values of .042 and .035, respectively). clades, yet did not fully support the internal relationships as RIBAS ET AL. | 921

FIGURE 2 Median-joining haplotype networks for (a) Cytb, 119 individuals, 1,050 bp (b), ND2, 116 individuals, 1,000 bp, (c) ND3, 120 individuals, 390 bp and (d) Fib5, 107 individuals, 505 bp [Colour figure can be viewed at wileyonlinelibrary.com] recovered by the mitochondrial gene tree (Figure S3.1 in the dates estimated by the species tree analysis (Figure 3). The high- Appendix S3). Dating analysis based only on the mtDNA data set est unidirectional migration rates were from R. melanosticta AtoR. using a gene tree approach in BEAST results in older dates when com- cristata and from R. melanosticta A to B. However, all HPDs for pared to those estimated by the species tree (Figure 3). The species migration estimates included zero, thus not excluding the possibility tree analysis including out-groups suggests an estimated date for the of no migration after divergence. The possible presence of migration origin of the Rhegmatorhina lineage at about 5 Ma (Figure S3.2 in between the two pairs of taxa analysed within the western clade Appendix S3). and the fact that IMa dating is congruent with the mtDNA dating Migration rates and divergence times were estimated under a suggests that migration after divergence may have affected the time coalescent framework for seven pairs of Rhegmatorhina species with estimates produced by the species tree analysis, as *BEAST does not adjacent distributions (Table S5). It was not possible to estimate consider the possibility of migration after isolation, considering all parameters between the two R. hoffmannsi lineages, probably due to similarities among species as ancestral polymorphism (Heled & their shallow divergence. Estimated divergence times fall within the Drummond, 2010). confidence intervals obtained in both the species tree and mtDNA dating analysis, but are in general more congruent with the mtDNA 3.5 | Historical demographic analysis dating. In the eastern clade, divergence between R. berlepschi and R. hoffmannsi A (point estimate at 0.5 Ma) was more recent, and migra- The EBSP analyses indicated increasing population sizes for most tion rates were higher than between R. berlepschi and R. gymnops species, with the exception of R. cristata and R. melanosticta A (Fig- (0.6 Ma), agreeing with the mtDNA topology and supporting a sister ure 4). There is evidence of gradual population growth in the last relationship between R. berlepschi and R. hoffmannsi. Within the 50,000–150,000 years, with R. berlepschi exhibiting the sharpest western clade, the divergence between R. cristata and R. melanosticta increase among all species (Figure 4). Rhegmatorhina cristata and R. B was estimated at about 1 Ma, whereas the divergence between melanosticta A showed no significant deviations of the null hypothe- the two R. melanosticta lineages was estimated at about 0.82 Ma; sis of neutrality and constant population for any marker, while R. both dates are in agreement with the mtDNA dating and older than berlepschi had significant deviations for all tests for both cyt b and 922 | RIBAS ET AL.

0.14 Ma 1 R. berlepschi 0.69 Ma 0.86 0.1 Ma 0.48 Ma/ 1 R. hoffmannsi B 0.25 Ma 0.84 Ma / 1 0.50 Ma 0.22 Ma R. hoffmannsi A 1 1

0.21 Ma R. gymnops 1 1.81 Ma / 1.60 Ma 1

0.19 Ma R. cristata 1

1.02 Ma / 0.61 Ma 0.24 Ma 1 R. melanosticta B 1 0.79 Ma / 0.36 Ma 0.97

0.17 Ma 1 R. melanosticta A

2.0 1.75 1.5 1.25 1.0 0.75 0.5 0.25 0.0

FIGURE 3 BEAST chronogram based on mtDNA. Values above branches are the mean of the node height (age) based on a calibration rate of 2.1% divergence per Myr for cytochrome b. Mean values obtained in the species tree analysis, when available, are shown below the values found on the BEAST analysis. Bars show the 95% HPD. Values below branches are posterior probabilities, shown only when larger than 0.95 [Colour figure can be viewed at wileyonlinelibrary.com]

ND2. The remaining species showed indications of expanding popu- to zero in both directions (Table S2.4 in Appendix S2). In addition, lations in some loci (Table S2.3 in Appendix S2). no hybrid specimens or specimens outside of the expected distribution ranges have been reported at the field or in scientific collections. This may indicate that although hybridization between R. 4 | DISCUSSION gymnops and R. hoffmannsi (A) occurs in the headwaters of the Tapajos River, it has had little effect on the phenotypes of these 4.1 | Distribution, diversity and taxonomy two species. Dense sampling throughout the distribution of Rhegmatorhina con- Our sampling of R. berlepschi extends south within the Madeira- firmed that the main Amazonian rivers delimit the geographical dis- Tapajos interfluve up to Jacareacanga (B4). Our northernmost sample tributions of most taxa as inferred from mtDNA lineages. of R. hoffmannsi is from the left margin of the Aripuana~ River (H2), Rhegmatorhina gymnops occupies a larger range than previously but Willis (1969) reported its occurrence as north as the Abacaxis reported (Zimmer & Isler, 2003), with the Juruena River as the west- River, a Madeira tributary, indicating that the two taxa occur at ern limit to its distribution (Figure 1). However, recent genomic anal- localities <200 km apart. More detailed sampling is needed to under- yses of samples from the southern limit of this distribution (at the stand this additional potential contact zone. Teles Pires River) have revealed a hybrid individual between R. gym- Most surprising, there are two distinct lineages of R. hoffmannsi nops and R. hoffmannsi (Weir et al., 2015). According to our mtDNA that overlap, with individuals from both lineages at the same locality topology, R. gymnops and the eastern lineage of R. hoffmannsi (A) are (H5) and in two adjacent localities close to Porto Velho (H1 and H8). not sister taxa and IMa analyses showed migration estimates close Denser sampling is needed to better understand the origin of this RIBAS ET AL. | 923

10 10

1 1 0.1 0.1 0.01 0.01 R. gymnops 0.001 0.001 R. berlepschii 0.0001 0.0001 1E-05

1E-05 1E-06

1E-06 1E-07 0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5

10 10

1 1

0.1 0.1

0.01 R. hoffmannsi A 0.01 R. hoffmannsi B

0.001 0.001

0.0001 0,0001

1E-05 1E-05

1E-06 1E-06 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0 0.1 0.2 0.3 0.4 0.5 10

0.1

0.01

0.001

0.0001 R. cristata

1E-05

1E-06

1E-07

1E-08

1E-09 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 10 10

1 1 0.1

0.01 0.1 0.001 R. melanosticta B

R. melanosticta A 0.0001 0.01 1E-05

0.001 1E-06

1E-07

0.0001 1E-08 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0 0.1 0.2 0.3 0.4 0.5

FIGURE 4 Extended Bayesian skyline plots (EBSPs) estimated in BEAST based on mtDNA and nuclear sequence data. Time scale in millions of years before present 924 | RIBAS ET AL. mtDNA structure, which may be related to major drainage capture Japura River is the distributional limit between R. melanosticta and R. events involving the Madeira River and its tributaries, as inferred cristata, with individuals of each species being captured at the south- previously for several avian lineages (Fernandes et al., 2012, 2014; ern (R. melanosticta) and northern (R. cristata) margins of this river in Sousa-Neves, Aleixo, & Sequeira, 2013; Ferreira et al., 2017; see also Vila Bittencourt, close to the Brazilian/Colombian border (M. Cohn- below). Detailed morphological analyses may help clarify whether Haft, personal communication). there is more than one diagnosable phenotype involved, but visual inspection of R. hoffmannsi skins at the MPEG and INPA collections 4.2 | Establishment of upland forests in western (about 70 specimens total) did not reveal any obvious difference in Amazonia plumage patterns (AA, CCR and M. Cohn-Haft personal observation). Two divergent mtDNA lineages are also evident within R. Rhegmatorhina and Gymnopithys species (Figures 1 and S3.2) occur melanosticta, with their distributional limits clearly associated with across the entire Amazon basin, Choco and Central America. The the Solimoes-Ucayali~ Rivers. The nuclear data set reveals two shared complex (G. leucaspis, G. bicolor, G. rufigula) is sister to Rhegmatorhina haplotypes between the two lineages and the dating analysis sug- with good support and includes altogether 12 recognized subspecies gests they have been isolated for about 0.8 Ma (Figure 3 and distributed in northern Amazonia, Choco and Central America. Table S2.4). The south-eastern mtDNA lineage (R. melanosticta B, Although G. bicolor has previously been considered a subspecies of M7-M22, Figure 1) corresponds to R. m. purusiana (type locality G. leucaspis, the G. leucaspis individual included in our analysis is sis- Cachoeira, on the Purus River, nearest to sampling locality M21; ter to G. rufigula with high support, indicating an expected closer Peters, 1951). Another previously recognized subspecies (R. m. badia; relationship between taxa that occur east of the Andes. Gymnopithys type locality La Pampa, nearest to M22) was recently merged with R. lunulata and G. salvini form the sister clade to the above group and m. purusiana (Zimmer & Isler, 2003), in agreement with our results. occur in western Amazonia (Napo and Inambari areas of endemism, Within the range of the north-western mtDNA lineage (R. melanos- respectively), probably separated by the same rivers (Solimoes-~ ticta A M1-M6, Figure 1) there are two described subspecies, sepa- Ucayali) that demarcate the distribution limits between R. melanos- rated by the Maranon~ River (R. m. melanosticta and R. m. ticta and R. purusiana. The Rhegmatorhina-Gymnopithys clade includes brunneiceps). We find only moderate support for this as samples at least six taxa from the western Amazonian areas of endemism from localities M3 and M5, corresponding to R. m. brunneiceps (type (Napo and Inambari). Both the distribution of Gymnopithys species, locality Moyobamba, nearest to M5; Peters, 1951), form a clade that mostly on the western portion of the Amazon basin, and the signifi- also includes one individual from M4, that has a phenotype typical cantly higher intraspecific diversity found in both cytb and ND2 for of R. m. melanosticta (RB personal observation) and occurs within its R. melanosticta (Table S2.3) suggest that the ancestor to Rhegma- range (north of the Maranon,~ type locality Sarayacu, nearest to M6). torhina had a western Amazonian distribution, occupying upland T. This result may suggest gene flow between R. m. melanosticta and R. Firme forests on the Andean slopes and possibly in western low- m. brunneiceps, despite noticeable phenotypic differences (crown col- lands. The fact that there is no Rhegmatorhina taxon east of the our, Zimmer & Isler, 2003), as has been found for several other spe- Negro and Xingu Rivers is consistent with a western ancestral distri- cies pairs separated by the Maranon~ River (Winger & Bates, 2015). bution of this genus, although extinction could produce the same Based on the evidence presented above, there are two distinct pattern. Taxonomic uncertainty and lack of sampling for Gymnopithys reciprocally monophyletic lineages within R. melanosticta correspond- preclude a biogeographic analysis of the whole group, but its initial ing to R. m. melanosticta and R. m. purusiana (the names with priority), diversification was probably affected by Pliocene landscape evolu- while the status of R. m. brunneiceps remains to be investigated in tion in western Amazonia. It is still debated in the literature when more detail. The fact that the morphologically more divergent R. m. upland (T. Firme) forest was established in western Amazonia. brunneiceps individuals group with R. m. melanosticta underscores a Although some authors think this may have happened towards the conflict between phenotypic and genetic traits (not observed in other end of the Miocene (Hoorn et al., 2010), the diversification history Rhegmatorhina taxa), and the comparatively higher migration rate esti- of this clade seems to support the alternative scenario of a dynamic mated from R. melanosticta A into R. melanosticta B suggests that landscape during the Plio–Pleistocene with more recent establish- introgression could have originated this pattern. Therefore, given this ment of this habitat (Latrubesse et al., 2010; Nogueira et al., 2013). conflict, we refrain from splitting R. melanosticta at this point, despite the reciprocal monophyly between R. melanosticta A and R. melanos- 4.3 | Diversification across the Madeira River ticta B, but recommend that future studies with denser sampling in terms of specimens and markers further investigate the issue, as there The first split within Rhegmatorhina separated Western Amazonian is clearly more than one evolutionary unit involved. (R. cristata and R. melanosticta) taxa from Brazilian shield taxa (R. ber- Rhegmatorhina cristata is a poorly known species that occurs in lepschi, R. gymnops and R. hoffmannsi). These clades are currently the recently described Jau area of endemism and is sparsely repre- separated by the Madeira River. The age of this split (about 1.6/ sented in biological collections. Our results show that it has high 1.8 Ma, Figures 3 and S3.1) coincides with the split associated with haplotype and nucleotide diversity and shares one nuclear haplotype this same river within Psophia (Psophia leucoptera versus Brazilian with both R. melanosticta lineages. Recent field data confirm that the shield clade), dated at about 1.6 Ma (CI 1.1–2.1) (Ribas et al., 2012). RIBAS ET AL. | 925

Splits between at least nine other clades of Amazonian T. Firme last 50,000 years (Figure 4). This result corroborates the suggestion understory birds (genera Willisornis, Schiffornis and Hylophylax, and of Weir et al. (2015) that a combination of climate change and river- complexes Myrmeciza hemimaelena, Automolus ochrolaemus, A. infus- ine barriers (River-Refuge ‘hypothesis’, Ayres & Clutton-Brock, 1992) catus, Sclerurus caudacutus, Xiphorhynchus pardalotus/ocellatus and could be related to diversification on opposite margins of the Thamnophilus aethiops) with distributions delimited by the Madeira Tapajos, with repeated cycles of isolation in drier periods and con- River also diverged from about 0.2 to about 2 Ma (d’Horta et al., tact in humid periods ultimately leading to speciation. The river- 2013; Fernandes et al., 2012, 2014; Smith et al., 2014; Sousa-Neves refuge mechanism predicts that barriers would be stronger in dry et al., 2013; Thom & Aleixo, 2015; de Deus Schultz et al., 2017) or periods due to the association between physical (rivers) and ecologi- have substantial genetic differentiation on opposite margins of the cal (open vegetation) factors. In addition, during dry periods, river Madeira River (Drymophila, Bates, Hackett, & Goerck, 1999). Despite discharge may have decreased (Govin et al., 2014), lowering the the variation in ages, it is clear that diversification within all these strength of the lower river courses as barriers to dispersal, but also groups of birds from the same geographic region and with distribu- originating less incised river channels and larger floodplains that also tions currently delimited by the same barrier (the Madeira River) constitute barriers for upland T. Firme birds. occurred during the last 2 Ma. This date corresponds quite well to The existence of three different lineages within the Rondonia^ area the history of the Madeira River itself. The origin of the Madeira of endemism (Madeira-Tapajos interfluve) highlights the complexity of River is probably related to the Pliocene orogeny of the Fitzcarrald this region already noted in previous phylogenetic studies (Fernandes arch, due to the subduction of the Nazca plate (Espurt et al., 2010). et al., 2014; Ribas et al., 2012). Willis (1969) commented on this com- The subduction started at about 4 Ma, and the uplift of the arch is plexity based on the distribution of Rhegmatorhina taxa within the subsequent to that. This uplift probably initiated a process that interfluve and attributed the parapatry of R. berlepschi and R. hoff- resulted in the connection of the lower Madeira with the Amazon mannsi to occasional changes in river directionality within the Madeira River in the upper Pliocene/Pleistocene. The differences among esti- and Tapajos basins. Here, we show that the pattern first described by mated ages for the splits in different groups may have several Willis is even more complex, involving three instead of two lineages. causes, including differential dispersal capacity of each group (Smith Latrubesse (2002) and Hayakawa and Rossetti (2015) report evidence et al., 2014), more recent changes in river course due to neotecton- for a recent establishment of the current course of the Aripuan~a River, ics (Hayakawa & Rossetti, 2015; Latrubesse, 2002), and/or recent describing palaeochannels that indicate that drainage was previously changes in river discharge due to changes in precipitation regimes distributed by other smaller rivers, such as the Manicore. The river through time (Govin et al., 2014; Wang et al., 2017). Although all capture and consequent establishment of the Aripuan~a are a possible the understory birds cited above occupy similar habitats, the few mechanism of recent isolation of understory bird populations. Our studies that have addressed their sensitivity to fragmentation have samples collected between the Aripuan~a and Manicore Rivers (from shown there are differences among them (Ferraz et al., 2007), which sites H2, H3 and H4) all group in the R. hoffmannsi A clade and are may have influenced the way each one responded to the formation identified as R. hoffmannsi. Interestingly, our dating analysis suggests or change in position of geographic barriers. that the split between R. hoffmannsi and R. berlepschi, which corresponds spatially to this river capture, occurred during the Pleistocene (about 0.7 Ma, Figure 3), as suggested by Latrubesse (2002). 4.4 | Contrasting effects of landscape evolution in The other river reported as being a barrier to understory bird dis- western and eastern Amazonia tributions within the Madeira-Tapajos interfluve is the Ji-Parana The three lineages from the western clade have their distributions River (Fernandes et al., 2014). The two R. hoffmannsi lineages seem delimited by the Japura and the Solimoes-Ucayali~ Rivers. The origin to be sympatric in portions of their ranges, as R. hoffmannsi A occurs of R. cristata at about 0.6 Ma and the split between the two R. on both margins of the Ji-Parana River, while R. hoffmannsi B was melanosticta lineages at about 0.4 Ma probably occurred in the not collected on the eastern margin. According to Hayakawa and absence of considerable changes in climate, considering that western Rossetti (2015), the Ji-Parana River drained northward and was Amazonia is probably less affected by drought in glacial periods located near 60 km eastward from its modern position, with several (Cheng et al., 2013) and that, consistent with this scenario, there is changes in its course in the recent past generating a radial palaeo- no signal of recent demographic change in R. cristata and R. melanos- drainage network. This dynamic history, associated with changes in ticta A (Figure 4, Table S4). This points to drainage evolution as one forest distributions during the Pleistocene, may have caused differ- possible driver of diversification and is in agreement with a Plio– ential isolation and contact among populations, and the scenario we Pleistocene establishment of upland forest in western Amazonia due describe here may be the result of the most recent changes in drai- to the entrenchment of the drainage system (Latrubesse et al., 2010; nage configuration and forest cover. Nogueira et al., 2013). Diversification in eastern Amazonia may have occurred under a 4.5 | Amazonian diversification in time and space different environmental regime, with marked periods of aridity in glacial times (Wang et al., 2017), reflected in the signal of recent popu- It is becoming clear in the literature that the major problem precluding lation expansion for all four lineages in the eastern clade during the a better understanding of diversification within Amazonia is trying to 926 | RIBAS ET AL. find single biogeographic explanations for the whole biome. Palaeocli- We are grateful to the efforts of many specimen collectors over many mate models indicate that Eastern and Western Amazonia follow dif- years of work in the Amazon Basin. C.C.R. and A.A. have permanent ferent climatic patterns, with a modest increase in precipitation in collecting permits issued by ICMBio/MMA (#20524-1 and 10420-1). western Amazonia and a significant drying in eastern Amazonia during Field and laboratory works were funded by the F.M. Chapman Fund/ the last glacial (Cheng et al., 2013; Wang et al., 2017). The geological AMNH, CNPq, FAPESP, CAPES and FAPEAM. C.G. was supported by history of these regions is also very distinct, with western Amazonia a joint post doctoral fellowship from CAPES/FAPEAM at PPG- apparently being covered by a large wetland system for a large portion GCBEV/INPA. C.C.R. and A.A. are supported by CNPq research pro- of the recent past, and eastern Amazonia being dominated by drainage ductivity fellowships (#308927/2016-8 and #310880/2012-2, respec- systems in the uplands of the Brazilian and Guiana shields (Hoorn tively). Invaluable support was also obtained through collaborative et al., 2010; Latrubesse et al., 2010; Nogueira et al., 2013). The mod- grants: Dimensions US-Biota-S~ao Paulo, co-funded by NSF (DEB els of drainage evolution suggest that the western and eastern drai- 1241066 to J.C.), NASA, and FAPESP (grant #2012/50260-6); PEER- nages were isolated during a large portion of their recent evolution USAID Cycle 5 to CCR, and NSF 1146423 to J.C. (Figueiredo, Hoorn, van der Vem, & Soares, 2009; Nogueira et al., 2013). In light of these differences, we should not expect a single pro- DATA ACCESSIBILITY cess to be the main driver of Amazonian diversification, and it is not realistic to accept or reject ‘universal’ hypotheses as the ‘Refugia All DNA sequences and alignments generated for this study are Hypothesis’ or the ‘Riverine Hypothesis’. available from GenBank (https://www.ncbi.nlm.nih.gov/genbank/). Our analysis of Rhegmatorhina exemplifies another instance of Pleistocene diversification in an Amazonian genus restricted to T. ORCID Firme understory forest. Based on broad sampling, a well-resolved taxonomy and strongly supported phylogenetic relationships, we Camila C. Ribas http://orcid.org/0000-0002-9088-4828 show that mitochondrial lineages corroborate the spatial patterns Alexandre Aleixo https://orcid.org/0000-0002-7816-9725 known for decades based on phenotypes of upland forest Amazo- Joel Cracraft http://orcid.org/0000-0001-7587-8342 nian birds in that they are geographically delimited by the large Ama- zonian rivers (Cracraft, 1985; Haffer, 1969). Our results also REFERENCES corroborate palaeoclimatic records obtained so far for the Amazon basin, with a signal of recent population expansion for all taxa, Ayres, J. M., & Clutton-Brock, T. H. (1992). River boundaries and species except for those from north-western Amazonia (R. cristata and R. range size in Amazonian Primates. American Naturalist, 140, 531–537. melanosticta) (Figure 4, Table S4), a region that was less affected by https://doi.org/10.1086/285427 Baker, P. A., Fritz, S. C., Dick, C. W., Eckert, A. J., Horton, B. K., Manzoni, past climatic oscillations (Cheng et al., 2013; Wang et al., 2017). S., ... Battisti, D. S. (2014). The emerging field of geogenomics: Con- These demographic changes are too recent to be related to taxic straining geologic problems with genetic data. Earth Science Reviews, origination, but this may indicate that southern and eastern Amazo- 135,38–47. https://doi.org/10.1016/j.earscirev.2014.04.001 € nian taxa associated with upland wet forests would also be suscepti- Bandelt, H.-J., Forster, P., & Rohl, A. (1999). Median-joining networks for inferring intraspecific phylogenies. Molecular Biology and Evolution, 16, ble to previous cycles of drier climate. The association between past 37–48. https://doi.org/10.1093/oxfordjournals.molbev.a026036 demographic changes related to palaeoclimatic cycles and the histor- Barrowclough, G. F., Cracraft, J., Klicka, J., & Zink, R. M. (2016). How many ically varying strength of rivers as barriers to dispersal may be the kinds of birds are there and why does it matter? PLoS ONE, 11(11), path to the answer to the long-standing question of which are the e0166307. https://doi.org/10.1371/journal.pone.0166307 Bates, J. M., Hackett, S. J., & Goerck, J. (1999). High levels of mitochon- drivers of Amazonian diversification. Instead of testing the ‘general drial DNA differentiation in two lineages of antbirds (Drymophila and ’ hypotheses of drainage evolution versus climate change, we should Hypocnemis). The Auk, 116, 1093–1106. https://doi.org/10.2307/ try to integrate both, spatially and temporally, and explicitly relate 4089688 them to biological diversification data. Borges, S. H., & Silva, J. M. C. (2011). A new area of endemism for Ama- zonian birds in the Rio Negro basin. The Wilson Journal of Ornithology, 124,15–23. ACKNOWLEDGEMENTS Brumfield, R. T. (2012). Inferring the origins of lowland Neotropical birds. The Auk, 129, 367–376. We thank the curator and curatorial assistants of the Academy of Nat- Cheng, H., Sinha, A., Cruz, F. W., Wang, X., Edwards, R. L., d’Horta, F. M., ... ural Sciences of Drexel University, Philadelphia, USA (ANSP); Field Auler, A. S. (2013). Climate change patterns in Amazonia and biodiversity. Nature Communications, 4, 1411. https://doi.org/10.1038/ Museum of Natural History, Chicago, USA (FMNH); Instituto Nacional ncomms2415 ^ de Pesquisas da Amazonia, Manaus, Brazil (INPA); Laboratorio de Gen- Cracraft, J. (1985). Historical biogeography and patterns of differentiation etica e Evolucßao~ Molecular de Aves, Universidade de S~ao Paulo, Brazil within the South American avifauna: Areas of endemism. Ornithologi- – (LGEMA); Lousiana State University Museum of Natural Science cal Monographs, 36,49 84. https://doi.org/10.2307/40168278 de Groot, R., Brander, L., van der Ploeg, S., Costanza, R., Bernard, F., (LSUMZ), Baton Rouge, USA; and Museu Paraense Emılio Goeldi, Braat, L., ... Hein, L. (2012). Global estimates of the value of ecosys- Belem, Brazil (MPEG), for loaning tissue samples. Mario Cohn-Haft tems and their services in monetary units. Ecosystem Services, 1,50– and Sergio H. Borges provided important information and discussion. 61. https://doi.org/10.1016/j.ecoser.2012.07.005 RIBAS ET AL. | 927 d’Horta, F. M., Cuervo, A. M., Ribas, C. C., Brumfield, R. T., & Miyaki, C. the divergence of Drosophila pseudoobscura and D. persimilis. Genetics, Y. (2013). Phylogeny and comparative phylogeography of Sclerurus 167, 747–760. https://doi.org/10.1534/genetics.103.024182 (Aves: Furnariidae) reveal constant and cryptic diversification in an Hoorn, C., Wesselingh, F. P., ter Steege, H., Bermudez, M. A., Mora, A., old radiation of rain forest understorey specialists. Journal of Biogeog- Sevink, J., ... Antonelli, A. (2010). Amazonia through time: Andean raphy, 40,37–49. https://doi.org/10.1111/j.1365-2699.2012.02760.x Uplift, climate change, landscape evolution, and biodiversity. Science, Drummond, A. J., & Rambaut, A. (2007). BEAST: Bayesian evolutionary 330, 927–931. https://doi.org/10.1126/science.1194585 analysis by sampling trees. BMC Evolutionary Biology, 7, 214. Hopkins, M. J. G. (2007). Modeling the known and unknown plant biodi- https://doi.org/10.1186/1471-2148-7-214 versity of the Amazon Basin. Journal of Biogeography, 34, 1400–1411. Espurt, N., Baby, P., Brusset, S., Roddaz, M., Hermoza, W., & Barbarand, https://doi.org/10.1111/j.1365-2699.2007.01737.x J. (2010). The Nazca Ridge and uplift of the Fitzcarrald Arch: Impli- Hung, C. M., Drovetski, S. V., & Zink, R. M. (2016). Matching loci sur- cations for regional geology in northern South America. In F. P. veyed to questions asked in phylogeography. Proceedings of the Hoorn & C. Wesselingh (Eds), Amazonia: Landscape and species evo- Royal Society B, 283, 20152340. https://doi.org/10.1098/rspb.2015. lution – a look into the past (pp. 89–100). Oxford: Blackwell Publ. 2340 Ltd.. Isler, M. L., Bravo, G. A., & Brumfield, R. T. (2014). Systematics of the Fernandes, A. M., Wink, M., & Aleixo, A. (2012). Phylogeography of the obligate ant-following clade of antbirds (Aves: Passeriformes: Chestnut-tailed Antbird (Myrmeciza hemimelaena) clarifies the role of Thamnophilidae). The Wilson Journal of Ornithology, 126, 635–648. rivers in Amazonian biogeography. Journal of Biogeography, 39, 1524– https://doi.org/10.1676/13-199.1 1535. https://doi.org/10.1111/j.1365-2699.2012.02712.x Lanfear, R., Calcott, B., Ho, S. Y. W., & Guindon, S. (2012). PartitionFin- Fernandes, A. M., Wink, M., Sardelli, C. H., & Aleixo, A. (2014). Multiple der: Combined selection of partitioning schemes and substitution speciation across the Andes and throughout Amazonia: The case of models for phylogenetic analyses. Molecular Biology and Evolution, 29, the spot-backed antbird species complex (Hylophylax naevius/Hylo- 1695–1701. https://doi.org/10.1093/molbev/mss020 phylax naevioides). Journal of Biogeography, 6, 1094–1104. https://doi. Latrubesse, E. (2002). Evidence of quaternary paleohydrological changes org/10.1111/jbi.12277 in middle Amazonia: The Aripuan~a-Roosevelt and Jiparana ‘‘fans’’. Ferraz, G., Nichols, J. D., Hines, J. E., Stouffer, P. C., Bierregaard, R. O. Jr, Zeitschrift Fur Geomorphologie, 129,61–72. & Lovejoy, T. E. (2007). A large-scale deforestation experiment: Latrubesse, E. M., Cozzuol, M., da Silva-Caminha, S. A. F., Rigsby, C. A., Effects of patch area and isolation on Amazon Birds. Science, 80, Absy, M. L., & Jaramillo, C. (2010). The late Miocene paleogeography 238–241. https://doi.org/10.1126/science.1133097 of the Amazon basin and the evolution of the Amazon river system. Ferreira, M., Aleixo, A., Ribas, C. C., & Santos, M. P. D. (2017). Biogeogra- Earth-Science Reviews, 99,99–124. https://doi.org/10.1016/j.earscire phy of the Neotropical genus Malacoptila (Aves: Bucconidae): The v.2010.02.005 influence of the Andean orogeny, Amazonian drainage evolution and Nogueira, A. C. R., Silveira, R., & Guimar~aes, J. T. F. (2013). Neogene- palaeoclimate. Journal of Biogeography, 44, 748–759. https://doi.org/ Quaternary sedimentary and paleovegetation history of the eastern 10.1111/jbi.12888 Solimoes~ Basin, central Amazon region. Journal of South American Figueiredo, J. J. P., Hoorn, C., van der Vem, P. H., & Soares, E. F. (2009). Earth Sciences, 46,89–99. https://doi.org/10.1016/j.jsames.2013.05. Late Miocene onset of the Amazon River and the Amazon deep-sea 004 fan: Evidence from the Foz do Amazonas Basin. Geology, 37, 619– Patel, S., Weckstein, J. D., Patane, J. S. L., Bates, J. M., & Aleixo, A. 622. https://doi.org/10.1130/G25567A.1 (2011). Temporal and spatial diversification of Pteroglossus Aracaris Govin, A., Chiessi, C. M., Zabel, M., Sawakuchi, A. O., Heslop, D., Horner,€ (Aves Ramphastidae) in the Neotropics: Constant rate of diversifica- T., ... Mulitza, S. (2014). Terrigenous input off northern South Amer- tion does not support an increase in radiation during the Pleistocene. ica driven by changes in Amazonian climate and the North Brazil Cur- Molecular Phylogenetics and Evolution, 58, 105–115. https://doi.org/ rent retroflection during the last 250 ka. Climate of the Past, 10, 10.1016/j.ympev.2010.10.016 843–862. https://doi.org/10.5194/cp-10-843-2014 Peters, J. L. (1951). Check-list of birds of the world, Vol. 7. Cambridge, Guindon, S., Dufayard, J. F., Lefort, V., Anisimova, M., Hordijk, W., & MA: Museum of Comparative Zoology, 318 pp. Gascuel, O. (2010). New algorithms and methods to estimate maxi- Rambaut, A., Suchard, M. A., Xie, D., & Drummond, A. J. (2014). Tracer mum-likelihood phylogenies: Assessing the performance of PhyML v1.6. Retrieved from http://beast.bio.ed.ac.uk/Tracer. 3.0. Systematic Biology, 59, 307–321. https://doi.org/10.1093/sysbio/ R Core Team (2014). R: A language and environment for statistical com- syq010 puting. R Foundation for Statistical Computing. Haffer, J. (1969). Speciation in Amazonian forest birds. Science, 165, Ribas, C. C., Aleixo, A., Nogueira, A. C. R., Miyaki, C. Y., & Cracraft, J. 131–137. https://doi.org/10.1126/science.165.3889.131 (2012). A paleobiogeographic model for biotic diversification within Harvey, M. G., & Brumfield, R. T. (2015). Genomic variation in a wide- Amazonia over the past three million years. Proceedings of the Royal spread Neotropical bird (Xenops minutus) reveals divergence, popula- Society of London Series B, 279, 681–689. https://doi.org/10.1098/ tion expansion, and gene flow. Molecular Phylogenetics and Evolution., rspb.2011.1120 83, 305–316. https://doi.org/10.1016/j.ympev.2014.10.023 Ribas, C. C., Joseph, L., & Miyaki, C. Y. (2006). Molecular systematics and Hayakawa, E. H., & Rossetti, D. F. (2015). Late quaternary dynamics in patterns of diversification of the Pyrrhura parakeets (Aves: Psittaci- the Madeira river basin, southern Amazonia (Brazil), as revealed by dae) with special reference to the P. picta/P. leucotis complex. Auk, paleomorphological analysis. Anais da Academia Brasileira de 123, 660–680. https://doi.org/10.1642/0004-8038(2006)123[660: Ci^encias, 87,29–49. https://doi.org/10.1590/0001- MSAPOD]2.0.CO;2 3765201520130506 Ronquist, F., Teslenko, M., van der Mark, P., Ayres, D., Darling, A., Heled, J., & Drummond, A. J. (2010). Bayesian inference of species trees Hohna,€ S., ... Huelsenbeck, J. P. (2012). MrBayes 3.2: Efficient Baye- from multilocus data. Molecular Biology and Evolution, 27, 570–580. sian Phylogenetic Inference and Model choice across a large model https://doi.org/10.1093/molbev/msp274 space. Systematic Biology, 61, 539–542. https://doi.org/10.1093/sysb Hey, J. (2010). Isolation with migration models for more than two popu- io/sys029 lations. Molecular Biology and Evolution, 27, 905–920. https://doi.org/ Rull, V. (2015). Pleistocene speciation is not refuge speciation. Journal of 10.1093/molbev/msp296 Biogeography, 42, 602–609. https://doi.org/10.1111/jbi.12440 Hey, J., & Nielsen, R. (2004). Multilocus methods for estimating popula- Schultz, E. D., Burney, C. W., Brumfield, R. T., Polo, E. M., Cracraft, J., & tion sizes, migration rates and divergence time, with applications to Ribas, C. C. (2017). Systematics and biogeography of the Automolus 928 | RIBAS ET AL.

infuscatus complex (Aves; Furnariidae): Cryptic diversity reveals west- Zimmer, K. J., & Isler, M. L. (2003). Family Thamnophilidae (Typical ant- ern Amazonia as the origin of a transcontinental radiation. Molecular birds). In J. del Hoyo, A. Elliot, & D. A. Christie (Eds), Handbook of the Phylogenetics and Evolution, 107, 503–515. https://doi.org/10.1016/j. birds of the world, Vol. 8 (pp. 448–681). Broadbills to Tapaculos Bar- ympev.2016.12.023 celona, ES: Lynx Editions. Silva, J. M. C., Rylands, A. B., & Fonseca, G. A. B. (2005). The fate of Zink, R. M., & Barrowclough, G. F. (2008). Mitochondrial DNA under Amazonian areas of endemism. Conservation Biology, 19, 689–694. siege in avian phylogeography. Molecular Ecology, 17, 2107–2121. Smith, B. T., McCormack, J. E., Cuervo, A. M., Hickerson, M. J., Aleixo, https://doi.org/10.1111/j.1365-294X.2008.03737.x A., Cadena, C. D., ... Brumfield, R. T. (2014). The drivers of tropical speciation. Nature, 515, 406–409. https://doi.org/10.1038/nature 13687 Sousa-Neves, T., Aleixo, A., & Sequeira, F. (2013). Cryptic patterns of BIOSKETCH diversification of a widespread Amazonian woodcreeper species complex (Aves: Dendrocolaptidae) inferred from multilocus phylogenetic The authors are broadly interested in the biogeography of the analysis: Implications for historical biogeography and taxonomy. Neotropics, with emphasis on Amazonian birds. This paper repre- – Molecular Phylogenetics and Evolution, 68, 410 424. https://doi.org/ sents work started by CCR as a F.M. Chapman Post Doctoral 10.1016/j.ympev.2013.04.018 Fellow at the American Museum of Natural History. Thom, G., & Aleixo, A. (2015). Cryptic speciation in the white-shouldered antshrike (Thamnophilus aethiops, Aves - Thamnophilidae): The tale of Author contributions: C.C.R. and J.C. conceived the study; a transcontinental radiation across rivers in lowland Amazonia and the northeastern Atlantic Forest. Molecular Phylogenetics and Evolu- C.C.R., F.M.H., R.B. and A.A. collected part of the samples; tion, 82,95–110. https://doi.org/10.1016/j.ympev.2014.09.023 C.C.R., C.G. and F.M.H. collected data and performed the analy- ... Wang, X., Edwards, R. L., Auler, A. S., Cheng, H., Kong, X., Wang, Y., ses; C.C.R., J.C. and C.G. wrote the manuscript with input from Chiang, H. W. (2017). Hydroclimate changes across the Amazon low- all other authors. lands over the past 45,000 years. Nature, 541, 204–207. https://doi. org/10.1038/nature20787 Weir, J. T., Faccio, M. S., Pulido-Santacruz, P., Barrera-Guzman, A. O., & Aleixo, A. (2015). Hybridization in headwater regions, and the role of rivers as drivers of speciation in Amazonian birds. Evolution, 69, SUPPORTING INFORMATION 1823–1834. https://doi.org/10.1111/evo.12696 Weir, J. T., & Schluter, D. (2008). Calibrating the avian molecular clock. Additional Supporting Information may be found online in the sup- Molecular Ecology, 17, 2321–2328. https://doi.org/10.1111/j.1365- porting information tab for this article. 294X.2008.03742.x Whitney, B. M., & Cohn-Haft, C. (2013). Fifteen new species of Amazo- nian birds. In J. del Hoyo, A. Elliott, & D. Christie (Eds), Handbook of the birds of the world, special volume: New species and global index (pp. How to cite this article: Ribas CC, Aleixo A, Gubili C, d’Horta – 225 239). Barcelona: Lynx Edicions. FM, Brumfield RT, Cracraft J. Biogeography and Willis, E. O. (1969). On the behavior of five species of Rhegmatorhina, diversification of Rhegmatorhina (Aves: Thamnophilidae): ant-following antbirds of the Amazon basin. Wilson Bulletin, 81, 363– 395. Implications for the evolution of Amazonian landscapes Winger, B. M., & Bates, J. M. (2015). The tempo of trait divergence in during the Quaternary. J Biogeogr. 2018;45:917–928. ~ geographic isolation: Avian speciation across the Maranon Valley of https://doi.org/10.1111/jbi.13169 Peru. Evolution, 69, 772–782. https://doi.org/10.1111/evo.12607