Phylogenetic relationships and biogeography of the ancient genus Onychorhynchus (Aves: Onychorhynchidae) suggest cryptic Amazonian diversity
Abstract
We examined phylogeographic patterns and cryptic diversity within the royal flycatcher, Onychorhynchus coronatus (Aves: Onychorhynchidae), a widespread Neotropical lowland forest tyrant flycatcher. A phylogeny of the six recognized subspecies was constructed from mtDNA sequence data of the NADH dehydrogenase subunit two gene, using Bayesian Inference and Maximum Likelihood methods. Phylogenetic analyses revealed high levels of intraspecific divergence within O. coronatus, supporting the existence of at least six independent lineages. The phylogenetic results uncovered the following relationships: (O. c. swainsoni [Southern Atlantic Forest], (O. c. coronatus [western Amazonia], (O. c. castelnaui [eastern Amazonia], (O. c. mexicanus [Central America], (O. c. occidentalis [Tumbesian], O. c. fraterculus [extreme northwestern South America])))). Biogeographic and dating analyses suggest that vicariant and dispersal events acted across approximately six million years to influence lineage diversification within this genus. Some of those events include the formation of the Amazon River and its tributaries, Andean uplift, and climatically induced vegetational shifts. Phylogenetic and biogeographic analyses of O. coronatus lineages support a hypothesis of area relationships in which the first divergence event isolated the Southern Atlantic Forest from Amazonia during the Late Miocene/Early Pliocene. This event was followed by the split of western and eastern Amazonia at the Early/Late Pliocene, the divergence of cis- and trans-Andean lowland regions also at the Early/Late Pliocene, the split between Central America and the extreme northwestern South America/Tumbes at the Early/Middle Pleistocene, and the split between extreme northwestern South America and Tumbes at Middle/Late Pleistocene. Subsequent divergence of the southern and northern populations in the western and eastern Onychorhynchus lineages took place during the Pleistocene. Comparison of phylogenetic trees and patterns in Onychorhynchus with those from published work suggests that across large New World radiations such as the Suboscines, some co-distributed lineages began to diverge long before others, which exemplifies the complexity of their evolutionary history.
Introduction
Neotropical forests harbor the world's highest biotic diversity, with birds being the most diverse and well-studied group of vertebrates in this region (Haffer 1969, Cracraft 1985, Brown 2014, Smith et al. 2014, Harvey et al. 2020). Despite multiple studies available, the origins of avian diversity throughout the Neotropics remain incompletely understood (Smith et al. 2014). Congruence of geographic ranges has allowed the identification of areas of endemism, which may represent common regions of biotic differentiation (Cracraft 1985). However, phylogenetic patterns of avian differentiation within the Neotropics, or the existence of spatial and temporal congruence of those patterns, continue to be explored (Cracraft and Prum 1988, Hackett 1993, 1996, Silva and Oren 1996, Bates et al. 1998, Weir and Price 2011, Ribas et al. 2012, Fernandes et al. 2014, Oliveira et al. 2017). In this context, generating organismal phylogenies help elucidate the patterns, mechanisms, and timing of species diversification in the Neotropical biota, shedding light on the region's evolutionary history (Harvey et al. 2020).
Molecular work conducted in the Neotropics during the last two decades has allowed the reconstruction of many phylogenies for different avian lineages, providing a better picture of the region's diversification history. Studies of widespread taxa show that phylogenetic patterns in Neotropical lowland birds vary across species, suggesting that even co-distributed avian lineages from different genera with distributions across Amazonia, the Andes, the Atlantic Forest, and Central America could have originated at different times and been influenced by various historical events (Smith et al. 2014). The history of the Neotropics includes complex events associated with the Andean uplift, the formation of riverine barriers and the Panamanian Isthmus, and vegetational changes caused by climatic shifts (Hoorn et al. 2010, Mora et al. 2010, O'Dea et al. 2016, Thom et al. 2020). Additionally, ecological and life-history differences among species, such as dispersal ability, may influence how they are affected by those barriers (Claramunt et al. 2012, Smith et al. 2014). This complexity in spatial and temporal patterns challenges previous expectations that a few large historical events or a single mechanism (e.g. vicariance) could alone underlie the profound diversity of birds found throughout the Neotropical lowlands (Leite and Rogers 2013, Naka and Brumfield 2018). Instead, species diversification is likely the product of multiple mechanisms acting together or at separate times and spaces (Smith et al. 2014). Thus, a more thorough understanding of the evolutionary history of the region requires the use of multiple phylogenies of broadly distributed taxa to uncover cryptic diversity and the complex patterns and processes behind this diversity, coupled with better models for the landscape history of the region (Bicudo et al. 2019).
The Neotropical landscape sets the stage where the diversity and evolutionary history of lowland forest birds developed. Amazonian rivers and the Andes are critical geographical features influencing the divergence of populations and the origin of taxonomic barriers for many widespread species in the Neotropics (Aleixo 2004, Cheviron et al. 2005, Miller et al. 2008, Milá et al. 2012, Oliveira et al. 2017). The Amazon basin is a vast lowland rainforest divided by a series of river courses, many draining to the Amazon River, which separates northwestern and southeastern Amazonia. The final uplift of the northern Andes during the Late Pliocene (5.0–2.7 million years ago [Mya], Gregory-Wodzicki 2000) resulted in the isolation of the extensive Amazon Basin from lowland forests west of the Andes. Also, the completion of the Panamanian Isthmus, approximately 3 Mya (Duque-Caro 1990, Coates and Obando 1996, Coates et al. 2004, Montes et al. 2015, Jaramillo et al. 2018), promoted the diversification of the Neotropical lowlands (Nores 2004, Smith and Klicka 2010, Johnson and Weckstein 2011). This land bridge connecting Central and South America enabled the exchange of biotas between these landmasses (Weir et al. 2009). The complex geography of the Neotropics, together with climatic and subsequent vegetation changes that took place throughout the Plio-Pleistocene, likely shaped the historical relationships of endemic areas and the spatiotemporal patterns of genetic variation for many lowland species (Aleixo 2004, Cheviron et al. 2005, DaCosta and Klicka 2008, Miller et al. 2008, Smith and Klicka, 2010, Milá et al. 2012, Ribas et al. 2012, Morrone 2014, Silva et al. 2019, Miranda et al. 2021). Therefore, phylogenetic studies with a comprehensive geographical and taxonomic sampling of lineages are crucial when investigating taxonomic differentiation and the historical relationships of areas of endemism in the Neotropics (Cicero et al. 2021).
Onychorhynchus coronatus (Aves: Onychorhynchidae) is distributed in the forested regions of continental America, from southern Mexico to southeastern Brazil, including extreme northwestern Peru and western Ecuador (Skutch 1960, Traylor 1979, Ridgely and Tudor 1994, Fitzpatrick et al. 2004). As such, it is an ideal candidate for studying biogeographic patterns across the Neotropical lowlands. This ancient lineage, whose origin dates to the Late Paleogene/Early Miocene (Ohlson et al. 2013, Harvey et al. 2020), is comprised of at least six disjunct populations distributed in major areas of endemism throughout the Neotropics (Fig. 1, Table 1). Resolving its historical biogeography and patterns of temporal diversification may provide important insights into understanding the relationships among areas of endemism and common factors influencing the diversification of other Neotropical lowland birds. Onychorhynchus coronatus inhabits humid and deciduous forests, mature secondary woodlands, and forest borders (Perrella et al. 2021), and is often observed foraging for insects in the understory (Traylor and Fitzpatrick 1982). Onychorhynchus coronatus has several distinctive characters, including a spectacular unique crest (rarely seen in the field), exceptionally long rictal bristles, large eyes, and unusually short legs (Fitzpatrick et al. 2004). Recent molecular studies of evolutionary relationships found that Onychorhynchus does not belong in the Tyrannidae (Tello and Bates 2007, Ohlson et al. 2008, Tello et al. 2009, Ohlson et al. 2013), and together with other former tyrannid genera, were placed in its own family, Onychorhynchidae (Ohlson et al. 2013, Harvey et al. 2020).
Taxon | Area of endemisma |
---|---|
swainsoni | Southern Atlantic Forest |
coronatus | Eastern Amazonia (Guiana, Negro, Belem, Tapajós, Xingú, Rondônia) |
castelnaui | Western Amazonia (Napo, Inambari) |
occidentalis | Tumbesian |
fraterculus | Extreme northwestern South America (Northern Choco, Nechí, Perijá, Santa Marta). |
mexicanus | Central America |
The genus Onychorhynchus has received different taxonomic treatments, including a single polytypic species with six subspecies (Traylor 1979, Fitzpatrick et al. 2004, Sample 2020, Clements et al. 2022), or four biological species (Cory and Hellmayr 1927, Pinto 1944, Wetmore 1972, Ridgely and Tudor 2009, BirdLife International 2023a, b2023b, c2023c, d2023d, Gill et al. 2023). Independently of the treatment followed, there is little doubt about the existence of six morphologically diagnosable allopatric taxa (Table 1, Fig. 1). Among these are two northern taxa, mexicanus (S and E Mexico to Panama) and fraterculus (N Colombia and NW Venezuela), which are often considered the ‘northern royal flycatcher'; two Amazonian forms, coronatus (SE Venezuela, the Guianas, and E Amazonian Brazil) and castelnaui (SE Colombia and S Venezuela S to E Ecuador, E Peru, W Amazonian Brazil and N Bolivia), the ‘Amazonian royal flycatcher'; a Tumbesian trans-Andean taxon, occidentalis (W Ecuador and extreme NW Peru), the ‘Pacific royal flycatcher'; and an Atlantic Forest form, swainsoni (E and SE Brazil, from Bahia and S Minas Gerais S to E São Paulo, E Paraná, and NE Santa Catarina), the ‘Atlantic royal flycatcher' (Sample et al. 2020).
Morphologically, these taxa vary in size, bill and tail lengths, extent of barring on breast, color of crest and rectrices (in the male), and brightness of the plumage, especially in the rump and tail (Fitzpatrick et al. 2004, Sample 2020, Fig. 1). Male coronatus have shiny scarlet crest feathers with a spot of black and steel-blue at tips, mostly dark brown upperparts, fine buff and black barring on lower back, cinnamon-rufous rump, tail chestnut-rufous (darkest towards tip), whitish throat, warm buff underparts, and fine brown barring across chest; castelnaui is very similar to coronatus, slightly smaller and with less barring above; swainsoni is paler generally, with bright ochraceous-buff underparts and no breast markings; mexicanus has the crest orange-red, rump and tail-coverts cinnamon-buff, tail tawny-orange shading to brownish at tip, underparts buffy yellow; fraterculus is very similar to mexicanus with fewer markings on breast and paler cinnamon rump and tail; and occidentalis is quite distinct, brighter and paler than swainsoni, upperparts warm tawny-brown, tail brighter and paler tawny (Fitzpatrick et al. 2004). Females, in general, differ from males in having the scarlet of the crest replaced with yellow or orange. Preliminary analyses of plumage coloration and morphometric measurements for some subspecies showed some differences, but the study did not include all taxa (Whittingham and Williams 2000). Other similarities in behavior and vocalization have been noted across subspecies (Ridgely and Tudor 1994, Fitzpatrick et al. 2004), but no formal review of the specific status of these subspecies has yet been undertaken.
Two of those forms, swainsoni from the Atlantic forests of southeastern Brazil and occidentalis from the Pacific forests of western Ecuador and northwestern Peru, represent small, endemic populations that face important conservation challenges (BirdLife International 2016, 2018), prompting some authors to propose elevating these taxa to full species rank (Collar et al. 1992, 1992, Wege and Long 1995, Stattersfield and Capper 2000, BirdLife International 2023b, d2023d).
In this study, we use mitochondrial DNA to: 1) propose a phylogenetic hypothesis within Onychorhynchus; 2) estimate divergence times of major lineages within this genus; and 3) interpret the historical relationships among areas of endemism using phylogenetic analysis and present-day patterns of species distributions.
Material and methods
Taxon sampling and data acquisition
A total of 40 O. coronatus samples representing all known taxa (mexicanus, fraterculus, coronatus, occidentalis, castelnaui and swainsoni) were analyzed to produce a comprehensive phylogeny of the genus. When available, we used multiple samples for each taxon to maximize the geographic coverage throughout the Neotropics and to allow the evaluation of genetic diversity within those taxa. DNA extraction, amplification and sequencing were conducted on samples of 25 O. coronatus specimens. Fifteen additional O. coronatus sequences and three outgroup (Terenotriccus erythrurus, Myiobius barbatus, and Iodopleura isabellae) sequences were obtained from GenBank (Supporting information). Terenotriccus and Myiobius are the closest relatives of Onychorhynchus within the tyrannid phylogeny (Tello et al. 2009, Ohlson et al. 2013, Harvey et al. 2020).
Genomic DNA was extracted from a small (~0.05 g wet weight) portion of tissue using the DNeasy tissue extraction kit following the manufacturer's directions (Qiagen, MD). The final pellet was resuspended in 200 μL of Qiagen DNA hydration solution. For blood samples (LBE, Supporting information), we used an in-house DNA extraction protocol (Peñafiel et al. 2019). For PCR amplification of the mitochondrial protein-coding gene, NADH dehydrogenase subunit two (ND2), we used the following primers: L5204 (Tello and Bates 2007); H6312 (Cicero and Johnson 2001); L5219 and H6313 (Sorenson et al. 1999). The ND2 gene is widely used in avian phylogeographic studies (Zink et al. 2006, Miller et al. 2008, Fernandes et al. 2012, Ritter et al. 2021) and there is strong indication that most mtDNA results can be taken as reliable indicators of population structure, phylogeographic patterns, and species limits (Zink and Barrowclough 2008).
Forward and reverse strands were sequenced on an ABI 3730xl Genetic Analyzer (Perkin Elmer, Foster City, California). Sequence assembly and editing were performed using Sequencher (ver. 4.1, Genecodes, Ann Arbor, Michigan, USA) and Geneious (www.geneious.com). All sequences generated in this study were deposited in GenBank (Supporting information).
Phylogenetic analyses
Phylogenetic relationships were studied using Bayesian inference (BI) and maximum likelihood (ML) analyses. We used jModelTest (Posada 2008) to choose the most appropriate model of DNA evolution, as determined by the Akaike information criterion (Posada and Buckley 2004). This analysis was conducted for the entire ND2 dataset, and four different partitions based on codon positions (1st, 2nd, 3rd and 1st 2nd).
We performed Bayesian analyses in MrBayes ver. 3.2 (Huelsenbeck and Ronquist 2001, Ronquist et al. 2012) using two independent runs of 10 million generations, the default temperature parameter, and default priors as starting values for the model parameters. Trees were sampled every 100 generations. Bayesian posterior probabilities were obtained from the 50% majority-rule consensus of all trees retained after a 10% burn-in. We used Tracer 1.6 (Rambaut et al. 2014) to ensure the log-likelihood scores converged on similar values past the burn-in phase. Clades with posterior probability values ≥ 0.95 were considered strongly supported. ML analyses were performed using the fast approximation ML algorithm in RaXML ver. 8.2.7 with 1000 bootstrap replicates (Stamatakis, 2006). Bootstrap percentage values were considered strong when ≥ 70%.
Additionally, we inferred intraspecific gene genealogies using statistical parsimony using the Program PopART ver. 1.5 (Leigh and Bryant 2015, http://popart.otago.ac.nz/index.shtml). Statistical parsimony networks have been used to assign specimens to species (Pons et al. 2006, Bond and Stockman 2008), corroborate known species, and aid in discovering new ones (Chen et al. 2010). To further explore the extent of lineage genetic differentiation, we calculated uncorrected pairwise percent sequence divergence among Onychorhynchus lineages.
Divergence time estimates
A likelihood-ratio test of the molecular clock hypothesis performed in Mega ver. 5.2 (Tamura et al. 2011) rejected the null hypothesis of equal evolutionary rate throughout the tree (p = 3.038 × 10−23). Thus, we used a relaxed-clock method to estimate the divergence times for the Onychorhynchus lineages.
Divergence time estimates were calculated using BEAST ver. 2.4.8 (Drummond and Rambaut 2007, Heled and Drummond 2010). Tree calibration was achieved using the ND2 substitution rate of 2.9% sequence divergence per million years (0.0145 substitutions/site/lineage/million years; Lerner et al. 2011). We used the coalescent Skyline tree prior (Drummond et al. 2005) instead of the Yule tree prior because of its better performance in estimating divergence times for datasets with independent branch rates and a mix of sequences from species and populations (Mello et al. 2020). The substitution rate was applied with a normal distribution on the uncorrelated log-normal relaxed molecular clock of the single partition. Two independent runs of 10 million generations were performed, sampling one tree every 1000 generations. We examined the marginal probabilities of all samples in Tracer 1.6 (Rambaut et al. 2014) to verify an effective sample size (ESS) exceeding 200 for all parameters. The two independent runs were combined using LogCombiner ver. 2.4.6 (Drummond and Rambaut 2007) after excluding a 10% burn-in from each run, and then a maximum-clade-credibility tree was obtained using TreeAnnotator ver. 2.4.6 (Drummond and Rambaut 2007). Intervals of divergence times were associated with their respective geologic periods following Gradstein et al. (2004). Divergence time estimates in the absence of any fossil calibration constraints may reduce the confidence of those estimates, thus those need to be used with caution (Warnock et al. 2015).
Biogeographic analysis
A Bayesian method of biogeographic and ancestral state reconstruction was applied in RASP (Yu et al. 2015), which implements Bayesian Binary MCMC (BBM). BBM offers a statistical procedure for inferring ancestral areas using a full hierarchical Bayesian approach (Ronquist 2004). For this analysis, species distribution areas were assigned to broad geographic regions. Here, five simplified regions were distinguished: A – Southern Atlantic Forest, B – Amazonia, C – Central America, D – Extreme Northwestern South America, and E – Tumbesian. The BBM analysis was run with the F81 + G model, the maximum number of areas in ancestral ranges was set to three to avoid broad distributions at the root node, and root distribution was set to outgroup. The analysis was run for one million generations using 10 chains, sampling every 100 generations. We discarded the first 25% as burn-in and combined two parallel runs into a final area tree.
Results
Phylogenetic relationships
In the BI analyses, a three-partition by-codon model (1st ND2 [GTR + I + G], 2nd ND2 [HKY + I] and 3rd ND2 [GTR + I + G]) had the best fit to the data. The BI majority rule tree recovered 13 nodes (excluding outgroup taxa), of which 12 had a ≥ 0.95 posterior probability support, providing a well-resolved phylogenetic hypothesis for Onychorhynchus relationships (Fig. 2). The ML tree was topologically identical to the BI tree and shared similar levels of node support (Fig. 2).
In the BI/ML trees, swainsoni is sister to a clade formed by mexicanus, occidentalis, fraterculus, castelnaui and coronatus. In this latter clade, occidentalis is sister to fraterculus, followed by mexicanus, castelnaui, and coronatus (Fig. 2). All nodes were strongly supported except for node (3) (unifying mexicanus, occidentalis, fraterculus, and castelnaui), which was weakly supported (0.51/46). The lack of support for this node suggests three possible lineage arrangements: that recovered by the BI/ML tree [coronatus, (castelnaui, (mexicanus, (occidentalis, fraterculus)))] and two alternative topologies: [(castelnaui, coronatus), (mexicanus, (occidentalis, fraterculus))] and [castelnaui, (coronatus, (mexicanus, (occidentalis, fraterculus)))], all representing permutations for the positioning of coronatus and castelnaui.
The 40 Onychorhynchus specimens comprised 21 unique haplotypes, and the statistical parsimony analysis revealed six distinct networks (Fig. 3). All six networks corresponded entirely and exclusively with described taxa: swainsoni, coronatus, castelnaui, occidentalis, fraterculus and mexicanus. The pairwise percent sequence divergence estimates showed that genetic distances between the described taxa ranges from 1.90 (fraterculus and occidentalis) to 10.98 (fraterculus and swainsoni) (Supporting information). The resulting tree and haplotype network also showed that the northern and southern Amazonian samples of castelnaui and coronatus are differentiated (Fig. 2, 3). However, given the small sample size for castelnaui and coronatus, further evidence needs to be explored to clarify whether these constitute two additional independent lineages.
Divergence time estimates and biogeographic analysis
According to the Bayesian ultrametric tree (Fig. 4), the separation of the ancestral Onychorhynchus lineage from its sister clade (Myiobius-Terenotriccus) occurred in the Late Paleogene to Early Miocene between 28.9 and 16.1 Mya (mean = 22.2 Mya). The age of the root of O. coronatus was placed in the Late Miocene to Early Pliocene between 7.7 and 4.6 Mya (mean = 6.1 Mya). Subsequent major splits within O. coronatus were estimated to have occurred between the Early Pliocene and Late Pleistocene (4.7–0.3 Mya). The first lineage to diverge was swainsoni, which separated in the Late Miocene to Early Pliocene between 7.7 and 4.6 Mya (mean = 6.1 Mya). Subsequent splitting took place in the Early to Late Pliocene between 4.7 and 2.9 Mya (mean = 3.8 Mya) and included the separation of coronatus from the ancestor of the mexicanus-occidentalis-fraterculus-castelnaui clade. Within this latter clade, castelnaui separated from the ancestor of the mexicanus-occidentalis-fraterculus clade in the Early to Late Pliocene between 4.2 and 2.5 Mya (mean = 3.3 Mya), mexicanus separated from the occidentalis-fraterculus clade in the Early to Middle Pleistocene between 1.4 and 0.7 Mya (mean = 1.0 Mya), and occidentalis separated from fraterculus in the Middle to Late Pleistocene between 1.0 and 0.4 Mya (mean = 0.7 Mya). Finally, the genetic distance between the northern and southern Amazonian samples of castelnaui and coronatus suggest that both experienced Northern–Southern Amazonia splits: one during the Early and Middle Pleistocene between 1.6 and 0.7 Mya (mean = 1.1 Mya) for castelnaui; and one during the Middle to Late Pleistocene between 0.8–0.3 Mya (mean = 0.6 Mya) for coronatus.
For the BBM ancestral area analysis, we used a reduced version of the BI/ML tree that included all the Onychorhynchus lineages (n = 8) and two outgroup taxa. In this analysis, the average standard deviation in split frequencies of Bayesian runs 1 and 2 was 0.002, which indicates the analysis reached convergence (Yu et al. 2015). A tree containing likelihood values of all ancestral areas at nodes (Fig. 4) identifies Amazonia (B) as the most likely area of origin for Onychorhynchus (B = 52.2%, node 1), followed by the Atlantic Forest (A = 26.1%), and a composite area including Amazonia and the Atlantic Forest (AB = 18.1%). Other middle nodes within Onychorhynchus were also found to have originated in Amazonia (node 2 [B = 95.6%] and node 3 [B = 96.1%]). The node (5) unifying the Central American mexicanus, extreme northwestern South America fraterculus, and Tumbesian occidentalis taxa, was reconstructed ambiguously as Central American (C) or Amazonian (B) (C = 42.9% vs B = 37.7%). Other alternative reconstructions at this node had less than 6% likelihood. Likewise, the node (9), unifying the extreme northwestern South America fraterculus and Tumbes occidentalis taxa, was reconstructed ambiguously as Extreme northwestern South American (D) or Tumbesian (E) (D = 42.8% vs E = 41.1%). Alternative reconstructions at this node had less than 8% likelihood.
Discussion
Spatiotemporal patterns of diversification within Onychorhynchus
The early divergence of the ancestral Onychorhynchus lineage from its sister clade Myiobius–Terenotriccus was estimated to occur in the Late Paleogene/Early Miocene (28.9–16.1 Mya) (Fig. 4). The existence of several small clades of ancient rainforest tyrannids (Onychorhynchus, Oxyruncus, Piprites, Neopipo) suggest a high persistence of lineages in this environment, but few opportunities for speciation early on (Ohlson et al. 2008). This outcome could result from more stable conditions of the extensive ‘pan-Amazonian' region during the Paleogene before the uplift of the Andes and the formation of the Amazon basin (Hoorn et al. 2010). The origin of the basal lineages in Onychorhynchus in the Late Miocene/Early Pliocene coincides with the timing of the major events (mountain building in the Central and Northern Andes, wetland progradation into Western Amazonia, restriction of ‘pan-Amazonia' by the uplift of the Northern Andes, megawetland disappearance, expansion of ‘terra firme' rainforests) that originated the Amazon rivers and its tributaries, which changed the connectivity of the Amazonian landscape (Hoorn et al. 2010).
The phylogenetic and biogeographic analyses support an Amazonian origin for Onychorhynchus during the Late Miocene/Early Pliocene (7.7–4.6 Mya), with an early expansion into the Atlantic Forest and a subsequent single colonization of Central America (Fig. 4). The spatiotemporal origins of Onychorhynchus lineages provide a snapshot of the complexity of lineage diversification that took place in the Neotropical lowlands. Three major geological events that have taken place in the recent biogeographical history of the Neotropical lowlands have influenced the diversification patterns in this taxon: the formation of the Andes, which caused the Amazon river drainage to change its course and helped to establish its current stage (Hoorn et al. 2010, Bicudo et al. 2019), the completion of the Panamanian Isthmus, connecting Central and North America with South America (Montes et al. 2015), and the series of Quaternary glacial-interglacial cycles (Hays et al. 1976, Rull et al. 2007). We discuss these events and their possible role in the diversification of Onychorhynchus below.
Relationships between Amazonia and the Southern Atlantic Forest
In Onychorhynchus, the current distribution of the Southern Atlantic Forest taxon, swainsoni, and its divergence time suggests a Late Miocene/Early Pliocene (7.7–4.6 Mya) southeastern route through the Chaco savannas as the expansion route from Amazonia. Previous bird studies have suggested two main routes for the historical dispersal of lineages between Amazonia and the Atlantic Forest (Batalha-Filho et al. 2012). The first is an older (Middle to Late Miocene, 23.0–5.6 Mya) southern route connecting southwestern Amazonia and Southern Atlantic Forest through the Paraná River basin, passing through the northern Chaco. The second is thought to be a younger (since the Early Pliocene, the last 5.5 Mya) route through northern Cerrado and Caatinga, with two other connection pathways in northeastern Brazil (one route that crosses the coastal zones of Maranhão, Piauí, Ceará, and Rio Grande do Norte and a second one that crosses Tocantins and Bahia) (Batalha-Filho et al. 2012). Other non-avian studies support these pathways (Por 1992, Costa 2003, Wang et al. 2004, Pellegrino et al. 2011). The divergence times we report are younger than those recently estimated for another distinct lineage of suboscines, Neopelma tyrant-manakins (Capurucho et al. 2018), in which divergence between Atlantic Forest and Amazonian taxa was estimated to be 18 Ma. This wide range of divergence times highlights a long history of connection and disconnection between these forested regions (Willis 1992).
Diversification within Amazonia
One important biogeographical pattern that emerges from the analysis of relationships within Onychorhynchus is the finding that the two Amazonian lineages (coronatus and castelnaui) are not sister taxa, a result also found in previous studies (Naka and Brumfield 2018). Within Amazonia, an ancient east/west pattern was found [(((occidentalis, fraterculus), mexicanus), castelnaui), coronatus], with the trans-Andean clade (occidentalis, fraterculus), mexicanus) being sister to the eastern Amazonian one castelnaui. Another interesting pattern within Amazonia was a more recent north/south genetic structure within castelnaui and coronatus.
The east/west Amazonian pattern
Our results suggest that the western Amazonian lineage castelnaui is sister to the trans-Andean mexicanus-occidentalis-fraterculus clade and not the eastern Amazonian lineage coronatus (node 3 in Fig. 2). Although weakly supported, this pattern makes sense biogeographically given the relatively close geographic distance separating the western Amazonian and trans-Andean clade and the presence of extensive river barriers (the Orinoco and Negro Rivers in northern Amazonia and the Madeira River in southern Amazonia) isolating the western and eastern Amazonian clades. Further genetic data is required to fully resolve this node.
Based on our results, the divergence between the western Amazonia/trans-Andean and eastern Amazonian clades occurred in the Early/Late Pliocene (4.7–2.9 Mya, Fig. 4). The disparity of time divergence estimates for a similar biogeographic pattern (Ribas et al. 2005, 2009, d'Horta et al. 2013, Smith et al. 2014) suggests that west/east splitting of Amazonian lineages has occurred independently at different times. Previous studies have shown that times of divergence across parapatric Amazonian populations tend to span throughout a wide range of periods, from the Late Miocene to the Late Pleistocene (8.5–0.2 Mya) (Naka and Brumfield 2018, Smith et al. 2014), suggesting that although common historical patterns may emerge, the general rule indicates idiosyncratic histories.
The east/west separation of the ancient Amazonian Onychorhynchus lineages took place in the Early/Late Pliocene times (4.7–2.9 Mya, Fig. 4), and probably originated from the various landscape changes that occurred in Amazonia since the late Miocene (Rossetti et al. 2005, Campbell et al. 2006, Figueiredo et al. 2009, Hoorn et al. 2010, d'Horta et al. 2013, van Soelen et al. 2017). Ancestral Amazonian Onychorhynchus lineages may have been distributed near the Andean foothills in the west and the Guianan/Brazilian shields in the east. Subsequent colonization of western Amazonian lowlands may have taken place after the extensive ‘Lake Pebas' transitioned to the modern Amazonian drainage pattern (Late Miocene, Rossetti et al. 2005, Hoorn et al 2010), and the modern upland ‘terra firme' forest became established in this area (Aleixo and Rossetti 2007). Out of all Amazonian areas, western Amazonia appears to be the most dynamic (Hoorn et al. 1995, Rossetti et al. 2005, Campbell et al. 2006, d'Horta et al. 2013, Smith et al. 2014), which is probably the result of the effects of the Andean uplift and a major drainage re-organization of the Amazon basin between the late Pliocene and early Pleistocene (Campbell et al. 2006; Espurt et al. 2010, Hoorn et al. 2010, Latrubesse et al. 2010, Mora et al. 2010, Bicudo et al. 2019). Crouch et al. (2018) used measures of species and phylogenetic diversity of passerine birds to independently assess that western Amazonia is underdispersed (i.e. in situ speciation determined) and composed of on-average younger lineages than eastern Amazonia. In contrast, the base of the Andes (and eastern Amazonia) is phylogenetically overdispersed (i.e. dispersal mediated) and comprised of on-average older lineages. These patterns are consistent with the history of Amazonian landscapes (Bicudo et al. 2019), showing that western Amazonian landscapes are younger, whereas eastern Amazonian ones are more ancient.
The north/south Amazonian pattern
The distribution of the Amazonian lineages in Onychorhynchus includes seven major areas of endemism: Napo and Inambari for castelnaui and Guiana, Negro, Belém, Tapajós, Xingú, and Rondônia for coronatus. In both lineages, our sampling allowed to uncover northern and southern splits that occurred in western castelnaui and eastern coronatus Amazonia at approximately 1.1 (1.6–0.7) and 0.6 (0.8–0.3) Mya, respectively (Fig. 4). Similar results have been found by an ongoing investigation of the genetic structure of Amazonian Onychorhynchus populations using UCEs (Luca T. Micheli pers. comm.). Similar published northern/southern Amazonian split estimates include those for Sclerurus guatemalensis (0.9 [1.2–0.6] Mya [western]) and S. mexicanus (0.7 [0.9–0.4] Mya [western]; 1.2 [1.5–0.9] Mya [eastern], d'Horta et al. 2013). Older published estimates include those for Pyrilia parrots (4.9 [5.2–4.6] Mya [western], 3.3 [3.4–3.2] Mya [eastern], Ribas et al. 2005). However, to further explore this pattern within castelnaui and coronatus, more widespread geographic sampling is needed in both lineages before concluding that these north and south populations represent independent lineages.
The origin of the trans-Andean clade
Andean mountain ranges are important contributors to the diversification of Neotropical lowland birds (Brumfield and Capparella 1996, Brumfield and Edwards 2007, Miller et al. 2008, Milá et al. 2012). Still, their effect as strong barriers has been debated across avian studies (Eberhard and Bermingham 2005, DaCosta and Klicka 2008, Ribas et al. 2009, Patel et al. 2011, d'Horta et al. 2013, Cadena et al. 2016, Hazzi et al. 2018).
The uncovered cis/trans Andean biogeographic pattern for Onychorhynchus highlights the importance of Andean uplift, directly or indirectly, in contributing to lowland diversification. Currently, the distribution of Onychorhynchus cis-(castelnaui) and trans-(occidentalis-fraterculus-mexicanus) Andean clades are bordered by the Eastern and Western Cordillera of the northern Andes, the Mérida Andes, and the Serranía de Perijá. The split between the trans-Andean clade and its Amazonian sister taxon occurred between 4.2 and 2.5 Mya (Fig. 4). The estimated divergence time for this split suggests two scenarios for the dispersal of the ancestor of the trans-Andean clade before or during the final uplift of the northern Andes between 2 and 5 Mya (Gregory-Wodzicki 2000). First, the uplift of the Andes served as a vicariance event driving the diversification of the clades on either side of the Andes (Graham et al. 2009, Patel et al. 2011, d'Horta et al. 2013, Fernandes et al. 2014). Second, dispersal during the uplift was likely followed by isolation and speciation; the dispersal may have occurred around the northern Andes through the Caribbean lowlands or low-lying passes through the Andes (Eberhard and Bermingham 2005, Miller et al. 2008, Ribas et al. 2009, Cadena et al. 2016). Several low-lying passes have been identified in the northern Andes across which avian lineages may have dispersed (Cadena et al. 2016). Given the closer geographic distance between the cis- and trans-Andean taxa, the Táchira Depression (extreme western Venezuela), Loja-Zamora Valleys (southern Ecuador), and Porculla Pass (northwestern Peru) are possible dispersal routes. However, other passes in central Colombia cannot be excluded. Discriminating between vicariance caused by Andean uplift and dispersal across the Andes based on temporal relationships is challenging, especially when factors like the dispersal abilities of the group have not been studied.
While some avian divergences across the Andes share similar age estimates with Onychorhynchus (DaCosta and Klicka 2008, Patané et al. 2009, d'Horta et al. 2013), others are older (Ribas et al. 2005, d'Horta et al. 2013) or younger (Ribas et al. 2007, 2009, Miller et al. 2008, Patané et al. 2009, Patel et al. 2011, Weir and Price 2011, Fernandes et al. 2014). These studies show the complexity of the origin of the cis/trans Andean lineages, and suggest that dispersal occurred before, during, and after the uplift of the northern Andes. Pre- and post-Andean uplift dispersal was probably facilitated by forest corridors that originated during Pliocene/Pleistocene interglacial periods (Chapman 1917, Haffer 1967). Some of these regions through which these putative forest corridors once passed are currently characterized by grassland and savannah ecosystems (Cardoso Da Silva and Bates 2002).
Tumbes-extreme northwestern South America–Central America relationships
The biogeographic analysis suggests that the ancestor of the occidentalis-fraterculus-mexicanus clade dispersed to Central America from northwestern South America during the Early Pliocene/Early Pleistocene (Fig. 4). The separation of the Western Andean fraterculus-occidentalis and Central American mexicanus clades occurred in the Early/Middle Pleistocene between 1.4 to 0.7 Mya (Fig. 4). This split was probably caused by the break between the lowland forests of northwestern Colombia and Panama during interglacial periods, when humid forests were replaced by drier vegetation, and the sea level raised 30–50 m (Haffer 1967, Brumfield and Capparella 1996, Nores 2004, 2020). This sequence of climatic changes suggests that the ancestor of the Western Andean clade fraterculus-occidentalis was distributed in extreme northwestern South America and then dispersed southward towards the Tumbesian region. Alternatively, if the dispersal across the Andes of the ancestor of the fraterculus-occidentalis-mexicanus clade took place through low-lying passes in southern Ecuador or northern Peru, the ancestor of fraterculus-occidentalis was probably distributed throughout western Ecuador and western Colombia. In both cases, the existence of a humid forest corridor along the lowlands west of the Andes may have allowed the expansion of this ancestral lineage along this area (Chapman 1926). During the expansion, the ancestor of the fraterculus-occidentalis clade probably retained its ancestral preferences toward humid habitats- at present, fraterculus is found in humid lowland forests- and eventually, occidentalis evolved to occupy drier and more seasonal habitats. Avian fossil evidence from northwestern Peru indicates that dry forest species had broader distributions during the transition between the last glacial and interglacial intervals (Oswald and Steadman 2015, Oswald et al. 2016). The phylogeographic break separating the extreme northwestern South American fraterculus and Tumbesian occidentalis lineages dates to the Pleistocene at approximately 0.7 (1.0–0.4) Mya. The separation of fraterculus and occidentalis may have resulted from geographical and ecological isolation caused by marine transgressions or parapatric speciation caused by extreme changes in habitat throughout the ancestral distribution (Nores 2004, 2020). The form occidentalis is endemic to the dry forests of the Tumbesian Centre of endemism in western Ecuador and northwestern Peru (Cracraft 1985, Best and Kessler 1995, Fig. 1). Of the Tumbesian endemic bird species whose presumed closest relatives are known, more than half have their closest relatives occurring in humid habitats, including Andean and lowland forests (Best and Kessler 1995, Dantas et al. 2016). Opposite to occidentalis, the extreme northwestern South American form fraterculus occurs in the understory of secondary and primary humid forests, especially along creeks of medium to small size. The distribution of this form is currently patchy, with populations in the humid forest belt from the lower Magdalena Valley to Choco (Magdalena-Urabá moist forest ecoregion, Dinerstein et al. 2017), in the eastern and western humid flanks of the Sierra Nevada de Santa Marta, and along the humid flanks of the Serranía del Perijá in the northern extreme of the Eastern Andes along the Colombia-Venezuela border.
Cryptic diversity in Onychorhynchus
The examination of mtDNA variation and its resulting tree suggest that Onychorhynchus is comprised of at least six independent lineages: mexicanus (Central America), occidentalis (Tumbes), fraterculus (extreme northwestern South America), castelnaui (western Amazonia), coronatus (eastern Amazonia), and swainsoni (Southern Atlantic Forest) (Fig. 2, 3). The form fraterculus, which is sometimes included within mexicanus, is genetically distinctive and with a distribution apparently delimited in northwestern Colombia by the Atrato River (Fig. 1). The mitochondrial tree also uncovered a shallow northern/southern Amazonian structure within castelnaui and coronatus (Fig. 2, 3), which would represent two additional lineages within Onychorhynchus. These distinctive lineages were not unexpected given the great amount of previously recognized geographic variation (subspecies) in Onychorhynchus, and the allopatric distribution of these forms (Traylor 1979). The monophyly of these lineages and their divergence times show that they have been evolving independently from 7.7 to 0.4 Mya (. Figs2–4).
The overall ND2 tree topology uncovered in this study agrees with Harvey et al. (2020) UCE tree, which is based on thousands of loci and included four of the six forms within the O. coronatus species complex. The stem and crown divergence date estimates for Onychorhynchidae (Onychorhynchus sister to Myiobius and Terenotriccus) and within Onychorhynchus also appear to be concordant with the estimates from Harvey et al. (2020).
Overall, marked patterns of genetic divergence, tree structure, and phenotypic differences within Onychorhynchus support the proposal that all six taxa most likely constitute independent lineages that deserve species status. Interestingly, all these forms, except fraterculus, were originally described as species (Traylor 1979), and four of these forms are recognized as species under some taxonomies (Gill et al. 2023), which further exemplifies their distinctive nature in terms of morphology and disjunct distribution.
Conclusions
Our study highlights the complex diversification patterns of the Neotropical lowlands and stresses the importance of assessing phylogenetic relationships of complex taxa with large and patchy distributions. Onychorhynchus constitutes an ancient radiation whose ancestral lineage originated in the extensive ‘pan-Amazonian' region in the Late Paleogene/Early Miocene. Onychorhynchus started radiating during the Late Miocene/Early Pliocene with an expansion out of Amazonia to the Atlantic Forest, followed by an Early/Late Pliocene splitting of an Amazonian ancestral lineage to form the western and eastern Amazonian clades, and a subsequent single Early/Late Pliocene trans-Andean colonization from western Amazonia that originated the ancestral trans-Andean clade. The trans-Andean lineages originated during the Early to Late Pleistocene.
Furthermore, this study helped to reveal independently evolving lineages that might have to be treated as separate species with different conservation concerns. Complementary studies that include nuclear DNA, morphology, niche differentiation, and vocalizations with thorough sampling throughout the Onychorhynchus distribution are needed to fully resolve the evolutionary relationships and delimit species within this genus.
Acknowledgements
– We thank the following museums and their curatorial and collection management staff for making tis-sue samples available for this study: Colección de tejidos Instituto de Investigación de Recursos Bio-lógicos Alexander von Humboldt; Academy of Natural Sciences of Philadelphia (Former Curator L. Joseph, N. Rice); American Museum of Natural History (J. Cracraft, G. Barrowclough, P. Sweet); Field Museum of Natural History (S. Hackett, D. Willard, T. Gnoske); and Louisiana State University Muse-um of Natural Science (R. T. Brumfield, F. Sheldon, Emeritus Curator J. V. Remsen, D. Dittmann). DNA work was carried out at the JGT Lab operated with support of the LIU-Brooklyn Biology Department, and the Ambrose Monell Molecular Laboratory from the Sackler Institute for Comparative Genomics at the American Museum of Natural History. Thanks to J. Morin, the LIU Department of Biology and Empresas Públicas de Medellín for providing funding for this study and J. Cracraft (AMNH) for allowing access to sequencing facilities at the AMNH. Thanks to Erin Sackett from the FMNH for help sequencing the swainsoni specimen, Gabriela Gavilanes for sequencing samples of occidentalis samples, D. Martínez, J. Llano, A. Chinome, A. Lopera and D. Beltrán for collecting samples of fraterculus, and Jefferson García-Loor and Amartya Mitra for collecting samples of occidentalis, and to Fundación Jocotoco for allowing us to conduct research at Reserva Buenaventura for the purpose of that collection. Collection and analyses of Ecuadorian samples were authorized by Contrato Marco de Acceso a Recursos Genéticos MAE-DNB-CM 2015-0017 granted by Ministerio del Ambiente to Universidad Tecnológica Indoamérica. We are grateful to J. Molina and T. Leslie for reviewing an early draft of the manuscript. We also thank two anonymous reviewers for their suggestions.
Funding
– This study was funded by the Department of Biology at Long Island University, Universidad San Francisco de Quito, Empresas Públicas de Medellín, and Universidad de Antioquia. LNN is currently supported by CNPq (Research Productivity Fellowship 310069/2021-1).
Permits
– Samples from live specimens were collected with permission from local authorities, voucher information for museum collected specimens are provided in the Supporting information.
Author contributions
Pamela Reyes: Conceptualization (equal); Formal analysis (equal); Methodology (equal); Writing – original draft (equal); John M. Bates: Formal analysis (equal); Methodology (equal); Writing – review and editing (equal); Luciano N. Naka: Formal analysis (equal); Methodology (equal); Writing – review and editing (equal); Matthew J. Miller: Formal analysis (equal); Methodology (equal); Writing – review and editing (equal). Isabel Caballero: Formal analysis (equal); Methodology (equal); Writing – review and editing (equal). Catalina González-Quevedo: Conceptualization (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Visualization (equal); Writing – review and editing (equal); Juan L. Parra: Conceptualization (equal); Methodology (equal); Writing – review and editing (equal); Hector F. Rivera-Gutierrez: Conceptualization (equal); Funding acquisition (supporting); Methodology (equal); Writing – review and editing (equal). Elisa Bonaccorso: Funding acquisition (equal); Methodology (equal); Writing – review and editing (equal). José G. Tello: Conceptualization (lead); Data curation (lead); Formal analysis (lead); Funding acquisition (lead); Investigation (lead); Methodology (equal); Visualization (equal); Writing – original draft (equal).
Open Research
Transparent peer review
The peer review history for this article is available at https://publons.com/publon/10.1111/jav.03159.
Data availability statement
Data are available from the Dryad Digital Repository: https://doi.org/10.5061/dryad.pvmcvdnrf (Reyes et al. 2023).