Time-Calibrated Molecular Phylogeny Reveals a Miocene–Pliocene

Organisms Diversity & Evolution (2018) 18:345–353 https://doi.org/10.1007/s13127-018-0373-7

ORIGINAL ARTICLE

Time-calibrated molecular phylogeny reveals a Miocene–Pliocene diversification in the Amazon miniature killifish genus Fluviphylax (Cyprinodontiformes: Cyprinodontoidei)

Pedro H. N. Bragança1 & Wilson J. E. M. Costa1

Received: 1 December 2017 /Accepted: 30 July 2018 /Published online: 15 August 2018 # Gesellschaft für Biologische Systematik 2018

Abstract Species of Fluviphylax are widely distributed over the Amazon and Orinoco river drainages and are among the smallest fish in the neotropics, inhabiting areas near the margin of slow-flowing clear and black water streams and lakes. Here, we present the first multigene molecular phylogeny of Fluviphylax, including all five nominal species of Fluviphylax and three undescribed species. The analysis included fragments of one mitochondrial and five nuclear genes, totaling 5880 bp. The dataset was analyzed using maximum parsimony, maximum likelihood, and Bayesian inference approaches providing high-supported well-solved trees. A time-calibrated analysis was performed providing information on the origin and diversification of the miniature genus in the Amazon. We estimate that Fluviphylax lineage splits from its sister group, the Anablepidae and Poeciliidae (Poeciliinae sensu Parenti, 1981), during the Late Eocene, about 36.6 Mya; but lineage diversification started only in the Middle Miocene, about 16 Mya, during the formation of the Pebas system. Subsequent splits within Fluviphylax occurred in the Late Miocene–Pliocene, between 10 and 6 Mya and during the Pliocene, and were probably influenced by paleogeographical events such as the breaching of the Purus arch, the rise of the Vaupés arch, the uplift of the Fitzcarrald arch, and the capture of the Contigo and Uraricoera river drainages by the Branco River. The present time-calibrated analysis provides the first insight on the evolution of one of the smallest vertebrate taxa in the Amazon and Orinoco river drainages.

Keywords Neotropical . Neogene . Paleodrainage . Pebas system . Amazon lampeye . Andean uplift

Introduction species richness has been attributed to the dynamic paleogeographical and paleoenvironmental events that have occurred The Amazon, with its complex fluvial system combining a in the area encompassing the Amazon and the neighboring mosaic of large rivers, streams, lakes, and palustrine wetlands, Orinoco River basin during the Neogene (Lundberg et al. is among the most species-rich biome in the world. This 1998;Hoornetal.2010; Wesselingh and Hoorn 2011). Miocene marine introgressions and the continuing uplift of the Andes are considered to have played an important role in Electronic supplementary material The online version of this article the evolution of the Amazon fish fauna, shaping its present (https://doi.org/10.1007/s13127-018-0373-7) contains supplementary diversity and distribution (Lundberg et al. 1998; Hoorn et al. material, which is available to authorized users. 2010). Sea level rise probably affected freshwater populations, isolating them in upland refuges that promoted allopatric spe- * Pedro H. N. Bragança [email protected] ciation, as well as allowing invasion and establishment of derived marine taxa along the present Amazon (Hubert and Wilson J. E. M. Costa Reno 2006; Lovejoy et al. 2006; Bloom and Lovejoy 2011). [email protected] Andean uplift during the Miocene and Pliocene leads to a complete change in the orientation and sediment inflow into 1 Laboratory of Systematics and Evolution of Teleost Fishes, Institute the Orinoco and Amazon River basins and has been also as- of Biology, Federal University of Rio de Janeiro, Caixa Postal 68049, Rio de Janeiro, CEP 21944-970, Brazil sociated with capture and isolation of river drainages (Hoorn 346 Bragança P.H.N., Costa W.J.E.M.

1993, 1994; Hoorn et al. 1995). According to fossil data and (Pollux et al. 2014;Pohletal.2015; Helmstetter et al. 2016; extant species distributions, the present Amazon fish fauna Reznick et al. 2017;Bragançaetal.2018). Recently, Bragança has been hypothesized to have appeared and diversified dur- et al. (2018) provided an updated classification for taxa previ- ing the Miocene, under the influence of those paleogeograph- ously grouped in the Poeciliidae: the African lampeyes were ical events (Lundberg et al. 1998; Lundberg et al. 2010). grouped in the Procatopodidae, Fluviphylax was recognized Recently, the development and improvement of more pre- as the only member of the Fluviphylacidae, and the cise fossil calibration techniques increased the role of time- Poeciliidae were restricted to the new world Poecillines. calibrated phylogenies in biogeographical studies In the present study, we provide the first multigene phylog- (Drummond et al. 2006; Donoghue and Benton 2007;Ho eny for Fluviphylax, comprising five nuclear and one mito- 2014). Divergence date estimates can lead to accurate recon- chondrial gene, including all five known nominal species and structions of the evolutionary past if analyzed under the light three still undescribed species. A fossil-calibrated analysis is of paleogeographical and paleoenvironmental data. Thus, fos- conducted, testing the influence of Amazon Neogene paleo- sil time-calibrated analysis of broadly distributed Amazon geographical events in the evolution and diversification of taxa can provide unique insight about the implications of Fluviphylax. Amazon landscape and drainage evolution on species distribution and diversification. The killifish genus Fluviphylax Whitley, 1965 is comprised of Materials and methods miniature species widely distributed over the Amazon and Orinoco river drainages, inhabiting areas near the margin of Taxon sampling slow-flowing clear and black water streams and lakes (Costa 1996; Costa and Le Bail 1999; Maldonado-Ocampo et al. Field studies in the Brazilian Amazon were conducted between 2008). Fluviphylax is generally found in small schools near the 2011 and 2016, when specimens of the five nominal species of surface associated with marginal vegetation or floating meadows, Fluviphylax and three undescribed species were collected. A list feeding on algae and small insects (Roberts 1970, 1972; of the examined material with their respective localities and Goulding et al. 1988). The small size, not surpassing 22 mm of collection number appears as Online Resource 1, and an updated standard length (SL), possibly prevents species of Fluviphylax distribution map of the genus is presented in Fig. 1. Molecular from being predated by large piscivorous fishes (Roberts 1972; data were obtained from specimens conserved in 98% ethanol Goulding et al. 1988;WeitzmanandVari1988). Five species and later preserved in the same fixative. In addition to these eight have been described for the Brazilian Amazon (Myers and species of Fluviphylax, 22 species representing all Carvalho 1955;Costa1996; Costa and Le Bail 1999), but recent Cyprinodontoidei families (sensu Parenti, 1981) were analyzed. field expeditions have revealed a previously unknown diversity, Outgroup comprised the aplocheiloid killifish Aplocheilus including records for the Orinoco River basin, providing a more lineatus (Valenciennes, 1846) (Aplocheilidae), in which the clear panorama on species distribution (Arrington and analyses were rooted. In order to avoid a large amount of miss- Winemiller 2003, 2006; Lasso et al. 2004; Maldonado-Ocampo ing data, sequences from closely related species were incorpo- et al. 2008; Montaña et al.2008; Bragança unpublished). rated to the analysis. Two sequences of Aplocheilus panchax Fluviphylax has been considered closely related to ovipa- (Hamilton, 1822) (KJ844835; KU692279) were included rous African lampeyes since it was first described, as a result among the data of A. lineatus (Valenciennes, 1846); one se- of similar patterns of fin position (Myers and Carvalho 1955; quence of Profundulus guatemalensis (Günther, 1866) Roberts 1970). Later, a morphology-based phylogenetic anal- (GQ119857) was included among the data of Profundulus ysis inferred the placement of both Fluviphylax and African labialis (Günther, 1866); one sequence of Crenichthys baileyi lampeyes (Aplocheilichthyinae) among the family (Gilbert, 1893) (AY356571) was included among the data of Poeciliidae, which also comprised the American live-bearing Crenichthys nevadae Hubbs, 1932; one sequence of Poeciliinae (Parenti 1981). Subsequent morphological phylo- Poeciliopsis occidentalis (Baird and Girard, 1853) genetic analyses not only supported the placement of (HQ556956) was included among the data of Poeciliopsis Fluviphylax and African lampeyes in Poeciliidae but also sug- elongata (Günther, 1866); and one sequence from Aphanius gested a closer relationship between them (Costa 1996; anatoliae (Leidenfrost, 1912) (KJ.552353.1) was included Ghedotti 2000). However, recent molecular analyses did not among the data of Aphanius isfahanensis Hrbek, Keivany, and support monophyly of Poeciliidae (sensu Parenti, 1981), indi- Coad, 2006. All sequenced specimens used in this study are cating Fluviphylax as an independent lineage closely related to deposited in the Ichthyological collection of the Institute of a new world clade comprising the Poecillinae and Biology, Federal University of Rio de Janeiro, Rio de Janeiro Anablepidae; whereas, the African lampeyes are supported (UFRJ) and in the Royal Museum for Central Africa, Tervuren to be more closely related to the cyprinodontoid old world (MRAC). A list of species with their respective localities and genera Aphanius Nardo, 1827 and Valencia Myers, 1928 GenBank accession numbers appears in Online Resource 2. Time-calibrated molecular phylogeny reveals a Miocene–Pliocene diversification in the Amazon miniature... 347

Fig. 1 Distribution map of Fluviphylax: F. obscurus (white dot), F. palikur (orange dot), F. pygmaeus (yellow dot), F. simplex (blue dot), F. zonatus (red dot), Fluviphylax sp. A (green dot), Fluviphylax sp. B (brown dot), and Fluviphylax sp. C (purple dot)

DNA extraction and sequencing Sequencing Mix (Applied Biosystems). Cycle sequencing reactions were performed in 20 μl reaction volumes containing The genomic DNA was extracted from muscle tissue of the 4 μlBigDye,2μl sequencing buffer 5 × (Applied right side of the caudal peduncle using a DNeasy Blood & Biosystems), 2 μl of the amplified products (10–40 ng), 2 μl Tissue Kit (Qiagen) according to the manufacturer’sinstruc- primer, and 10 μl deionized water. The thermocycling profile tions. PCR amplification was performed using the following was as follows: (1) 35 cycles of 10 s at 96 °C, 5 s at 54 °C, and primers: Glyt_F577 and Glyt_R1562 (Li et al. 2007) for the 4 min at 60 °C. The sequencing reactions were purified and glycosyltransferase gene (GLYT1); myh6_F507 and denatured, and the samples were run on an ABI 3130 Genetic myh6_R1325 (Li et al. 2007) for the cardiac protein- Analyzer. Sequences were edited using MEGA 6 (Tamura et encoding gene, myosin heavy chain 6 (MYH6); al. 2013), in which both the forward and reverse sequences SREB2_F10 and SREB2_R1082 (Li et al. 2007)forthe were analyzed and used to build the final sequence. Among super-conserved receptor expressed in brain 2 gene the nuclear genes, heterozygous sites were completed by its (SREB2); RAG1F1 (López et al. 2004) and H3405 (Hrbek correspondent-degenerated nucleotide code. Sequences were et al. 2007) for the recombination activation gene 1 (RAG1); aligned using ClustalW (Chenna et al. 2003). The DNA se- ENC1_F88 and ENC1_R975 (Li et al. 2007) for the ectoder- quences were translated into amino acids residues to test for mal neural cortex 1 gene; and LCO1490 and HCO2198 the absence of premature stop codons or indels using the pro- (Folmer et al. 1994) for the cytochrome c oxidase subunit I gram MEGA 6.0. Alignment data is present in Online (COI), totalizing 5880 bp. Polymerase chain reactions (PCR) Resource 3. were performed in 30 μl reaction mixtures containing 5 × Green GoTaq Reaction Buffer (Promega), 3.6 mM MgCl2, Phylogenetic analysis 1 μM of each primer, 50 ng of total genomic DNA, 0.2 mM of each dNTP, and 1 U of Taq polymerase. The thermocycling The dataset was partitioned according to each gene. The best- profile was as follows: (1) 1 cycle of 4 min at 94 °C; (2) fit evolutionary model was calculated for each partition using 35 cycles of 1 min at 92 °C, 1 min at 49–60 °C (varying the Akaike Information Criterion (AIC) determined by the according to the primer and the sample), and 1 min at 72 °C; jModelTest 2.1.7 (Darriba et al. 2012). The best model for and (3) 1 cycle of 4 min at 72 °C. In all PCR reactions, neg- each partition is present in Online Resource 4. To check for ative controls without DNA were used to check contamina- major discordance among individual gene trees, maximum tions. Amplified PCR products were purified using the Wizard likelihood trees were generated for each gene alignment, using SV Gel and PCR Clean-Up System (Promega). Sequencing Garli 2.0 (Zwickl 2006). The phylogenetic analyses were con- reactions were made using the BigDye Terminator Cycle ducted through Bayesian inference (BI), using the program 348 Bragança P.H.N., Costa W.J.E.M.

MrBayes v3.2.5 (Ronquist et al. 2011) and Beast v.1.8 supported in all analyses, presenting maximum bootstrap (Drummond et al. 2012); maximum likelihood (ML), using (MP = 100; ML = 100) and maximum posterior probability the program Garli 2.0 (Zwickl 2006); and maximum parsimo- values (BI = 1). In all analyses, a clade comprising ny (MP) using PAUP4 (Swofford 2003). When performing Fluviphylax, the Poeciliidae (sensu Rosen and Bailey 1963; MrBayes v3.2.5, BI was conducted using two Markov chain Poeciliinae sensu Parenti 1981) and Anablepidae (sensu Monte Carlo (MCMC) runs of two chains each for 3 million Parenti 1981) is well-corroborated, with high support values generations, a sampling frequency of 1000; and in Beast v.1.8 (Fig. 2 and Online Resources 5 and 6). Fluviphylax was sup- (Drummond et al. 2012), the BI was performed with 30 mil- ported to be the sister group of the clade containing lion generations, with a sampling frequency of 1000. The Poeciliidae and Anablepidae with high bootstrap value in the quality of the MCMC chains was evaluated in Tracer 1.6 ML (81) and maximum posterior probability value in the BI (Rambaut et al. 2013), and a 25% burn-in was removed in (1). However, in the MP analysis, it was not possible to define TreeAnnotator v1.8.4 (Drummond et al. 2012). The values relationships between the three lineages (Online Resouce 6). of support of the ML and MP analyses were calculated by A sister group relationship between the African lampeyes, 1000 bootstrap replications (Felsenstein 1985). Valenciidae, and Aphaniidae was also highly supported in all analyses (Fig. 2 and Online Resources 5 and 6). The clade Time-calibrated tree including Fluviphylax, Poeciliidae, and Anablepidae is supported as sister to the clade comprising the African lampeyes, The divergence time analysis was performed in Beast v.1.8 the Valenciidae, and the Aphaniidae. (Drummond et al. 2012), using the concatenated dataset in- In all the three analyses, relationships among species of cluding the same partitions as described above, and an uncor- Fluviphylax presented the same topology, supported by high related relaxed clock model, which emphasizes the minimum bootstrap and posterior probability values. Fluviphylax age and has been considered appropriate for fossil calibration palikur Costa and Le Bail, 1999 was indicated to be sister to points (Parham et al. 2012). Bayesian inference was per- a clade comprising all other congeners, with maximum sup- formed with the following settings: 30 million generations port values, and Fluviphylax sp. A, sister to a clade including with a sampling frequency of 1000. The value of parameters the remaining species of the genus, which included two main of the analyses, convergence of the MCMC chains, sample well-supported lineages, one including Fluviphylax pygmaeus size, and the stationary phase of chains were evaluated using Costa, 1996 and Fluviphylax simplex Costa, 1996 and the Tracer 1.6 (Rambaut et al. 2013). A Yule process for the tree other, Fluviphylax obscurus Costa, 1996, Fluviphylax sp. B, prior (Gernhard 2008) was used, establishing two calibration Fluviphylax zonatus Costa, 1996, and Fluviphylax sp. C. points. Fossil choice for calibration followed Costa et al. Maximum likelihood and BI analyses including only nucle- (2017), where the fossil Prolebias stenoura Sauvage, 1874, ar partitions and analyses with only the mitochondrial gene was placed at the stem comprising the genera Aphanius, COI were also performed for checking eventual discordance Valencia, and the African lampeyes, corresponding to the or- between mitochondrial and nuclear data (Online Resources 7 igin of the crown old world cyprinodontoid clade (Pohl et al. and 8). The maximum likelihood and Bayesian inference anal- 2015), estimated to have occurred at least 33 Mya (prior set- ysis based on the nuclear partitions resulted in the same topol- ting: exponential distribution, mean = 33, and standard devia- ogy, identical to ML and BI topologies including both nuclear tion = 0.5), and the fossil Aphanius chebianus (Obrhelová, and mitochondrial data, but some nodes received slightly lower 1985), placed at the node where Aphanius and Valencia di- support values (Online Resource 7). However, most clades verge, corresponding to the most ancient record of Aphanius, recovered in the ML and BI analysis including only COI re- with an estimated age of 17 Mya (prior setting: exponential ceived extremely low support values, and some clades were distribution, mean = 17, and standard deviation = 0.5). recovered as paraphyletic (e.g., Anablepidae, Procatopodidae). Fluviphylax internal nodes were little supported and in contrast to the analysis including both nuclear and mitochondrial data Results and the analysis comprising only the nuclear partitions, Fluviphylax sp. A was considered to be the sister species to Phylogenetic relationships Fluviphylax zonatus and Fluviphylax sp. B, instead of being sister to all other Fluviphylax species except F. palikur (Online The ML and the BI analyses presented the same topology Resource 8). (Fig. 2 and Online Resource 5 for the BI performed in MrBayes v3.2.5), but in the MP (Online Resource 6), it was Divergence dates not possible to establish the relationships between the deeper lineages of Cyprinodontoidei. However, monophyly of the Divergence time estimates indicate that Fluviphylax diverged clade comprising all species of Fluviphylax was strongly from its sister group comprising the Poeciliidae and the Time-calibrated molecular phylogeny reveals a Miocene–Pliocene diversification in the Amazon miniature... 349

Fig. 2 Phylogenetic relationships among 30 species of Cyprinodontoidei, probability values derived from the Bayesian analysis (Beast v.1.8), num- including all nominal species of Fluviphylax and three undescribed bers in the right are bootstrap support values above 50% taken from the species, inferred by using partial sequences of the nuclear-encoded genes maximum likelihood analysis. Asterisk means maximum values, and GLYT1, ENC1, RAG1, MYH6,andSREB2 and the mitochondrial gene colors under species name are related to the color of each species on COI, total of 5880 bp. Numbers left to the bar indicate posterior distribution map (Fig. 1)

Anablepidae in the Late Eocene (36.6 Mya; 95% HPD Discussion [highest posterior distribution], 31.9–42.5 Mya) (Fig. 3). However, lineage diversification started in the Middle Phylogenetic position and relationships Miocene, with the split of F. palikur (16.3 Mya; 95% HPD, among species of Fluviphylax 11.4–21.8 Mya). Subsequent splits within Fluviphylax occurred in the Late Miocene–Pliocene, between 10 and 6 The present analysis is consistent with recently published mo- Mya, and in the Pleistocene, around 1 Mya (Fig. 3). lecular phylogenies in which the monophyly of the Poeciliidae Fluviphylax sp. A lineage diverged during the transition be- (sensu Parenti 1981) is refuted (Pollux et al. 2014;Pohletal. tween the Middle and Late Miocene (10.1 Mya; 95% HPD, 7– 2015; Helmstetter et al. 2016; Reznick et al. 2017; Bragança et 13.4 Mya). Around the same period, two main lineages, one al. 2018). Despite the morphological resemblance between with F. simplex and F. pygmaeus and the other with F. zonatus, Fluviphylax and the African lampeyes, Fluviphylax is highly Fluviphylax sp. B, F. obscurus,andFluviphylax sp.C,also supported as being more closely related to the American live- splitted (8.7 Mya; 95% HPD, 6.1–11.9 Mya). Fluviphylax bearers (Poeciliidae) and to the Anablepidae, whereas the pygmaeus and F. simplex diverged in the transition between African lampeyes are indicated to be more closely related the Late Miocene and the Pliocene (6.5 Mya; 95% HPD, 3.9– to the cyprinodontoid old world genera Aphanius Nardo, 9.4 Mya), and around the same period the Negro River spe- 1827 and Valencia Myers, 1928 (Pollux et al. 2014;Pohlet cies, F. zonatus, F. obscurus, Fluviphylax sp.B,and al. 2015; Helmstetter et al. 2016; Reznick et al. 2017; Fluviphylax sp. C, split in two distinct lineages (6.3 Mya; Bragança et al. 2018). However, in contrast to previous 95% HPD, 4.0–9.0 Mya), one with F. zonatus and studies where the placement of Fluviphylax as the sister Fluviphylax sp. B and the other with F. obscurus and group of a clade comprising the Poeciliidae and the Fluviphylax sp. C. Fluviphylax zonatus and Fluviphylax sp. Anablepidae relied on considerably low support values B diverged during the Pleistocene (1.0 Mya; 95% HPD, 0.4– (Pollux et al. 2014; Helmstetter et al. 2016), all analyses 1.8 Mya), as well as F. obscurus and Fluviphylax sp. C (1.0 of the present study corroborated this phylogenetic posi- Mya; 95% HPD, 0.4–1.8 Mya). tion with high support values. 350 Bragança P.H.N., Costa W.J.E.M.

Fig. 3 Time-calibrated phylogeny of the Cyprinodontoidei and calibrated by the age of cyprinodontiform fossil record, and colors under Fluviphylax obtained from the Bayesian dating analysis in Beast v.1.8. species name are related to the color of each species on distribution map Bars represent maximum and minimum date estimates for each node, and (Fig. 1) the numbers are node divergence mean ages. Arrows indicate the nodes

This study corroborates previous morphological studies F. zonatus is considered to be closely related to the that indicated F. palikur as sister to a clade containing all other undescribed Fluviphylax sp. B. congeners (Costa and Le Bail 1999; Bragança unpublished). The undescribed species Fluviphylax sp. A is herein consid- Fluviphylax diversification and paleogeographical ered to be sister to a clade comprising all other congeners evolution of northern South America during Miocene except F. palikur, which is congruent with ongoing studies and Pliocene using external morphology and osteology of Fluviphylax sp. A (Bragança, unpublished). The remaining species of Our time-calibrated analysis indicated that the ancestor of the Fluviphylax are grouped in two lineages, one with F. Fluviphylax lineage diverged from its sister group, a clade pygmaeus and F. simplex and another with F. obscurus, F. comprising anablepids and poeciliids in the Eocene, at about zonatus, Fluviphylax sp.B,andFluviphylax sp. C. A sister 37 Mya (Fig. 3). Considering that basal lineages of anablepids species relationship between F. pygmaeus and F. simplex was (e.g., genera Oxyzygonectes and Anableps) and poeciliids firstly proposed by Costa (1996) and is also supported in the (e.g., Tomeurus and Phalloptychus) are uniquely found in es- present analysis. Fluviphylax obscurus is considered to be tuarine brackish waters of tropical America (Ghedotti 1998; closely related to the undescribed Fluviphylax sp. C, whereas Lucinda 2003), a common condition among new world Time-calibrated molecular phylogeny reveals a Miocene–Pliocene diversification in the Amazon miniature... 351 cyprinodontoids (e.g., Rosen 1973), we assume that the an- Negro drainages may be explained by a former isolation in cestor of Fluviphylax lived in similar habitats. Although there the Orinoco basin just after the rise of the Vaupés arch, follow- are no available records for Fluviphylax species inhabiting ed by recent dispersal to the upper Negro drainage as a con- brackish water, the distribution of F. palikur, the most basal sequence of the present low topographic relief between the species of the genus, along coastal river basins (Fig. 1)maybe Orinoco and Negro watersheds, which have allowed ephem- an indicative of this past brackish water tolerance. eral and year-round connections, thus enabling gene flow be- In contrast to F. palikur, the clade comprising all other tween fish populations from the Negro and Orinoco river species of Fluviphylax is distributed in the western-central drainages (Winemiller et al. 2008; Winemiller and Willis Amazon River basin and in the Orinoco River basin (Fig. 1). 2011). This distribution pattern is geographically partially concordant The capture of the upper Negro river drainage by the with the Miocene Pebas lake-wetland system (about 24–11 Amazon River allowed the subsequent dispersion of Mya), which was fed by Andean rivers in the west, small Fluviphylax over central Amazon River basin. The continuous cratonic drainages in the east, and connected to the sea in Andean uplift led to a change in the amount of Andean sedi- the present Venezuela Caribbean coast (e.g., Wesselingh and ments carried by the Amazon River, and around 7 Mya sedi- Salo 2006; Wesselingh and Hoorn 2011). Based on paleonto- mentation rates in the Amazon fan increased (Figueiredo logical pore and spore data, Hoorn (1993, 1994) has found 2009, 2010). Probably, Fluviphylax dispersed over the evidence that the Pebas system was mostly a freshwater envi- Amazon basin before the increase in sediment rate transport ronment, affected by some episodically marine introgression and between the series of periodic marine transgressions events. More recently, Jaramillo et al. (2017) described two events that happened in the Late Miocene. According to our main introgression events in the Pebas system, one in the date estimates, the increase in the sediment load over the Early Miocene between 18.1 and 17.2 Mya and the other in Amazon-Solimões River and the marine transgression events the Middle Miocene between 16.1 and 12.4 Mya. Therefore, resulted in the initial diversification of the clade comprising the divergence timing analysis indicating a split between F. species endemic to central Amazon around 9 Mya. Later, palikur and the Amazon-Orinoco clade at about 16 Mya sup- about 6.5 Mya, the increase in the sediments outflow in the ports the hypothesis that the ancestor of the Amazon-Orinoco Solimões and in the Madeira River basin resulted in the split Fluviphylax clade colonized the Pebas system during a Middle between F. pygmaeus, endemic to the Madeira river drainage, Miocene marine introgression event. and F. simplex, widespread along Solimões-Amazon flood- According to our analysis, species diversification within plains. This disjunction may have been favored by the subse- the Amazon-Orinoco Fluviphylax clade took place between quent uplift of the Fitzcarrald ridge about 4 Mya, a major the Late Miocene (about 10 Mya) and Pleistocene (about 1 event in southwestern Amazon, responsible for compartmen- Mya). This timing is congruent with the establishment of the talization of Solimões-Amazon and Madeira drainages modern Amazon River basin configuration, as a consequence (Espurt et al. 2007;Espurtetal.2010). of continuing uplift of the Eastern Cordillera in the Late In the Pliocene, around 5–4 Mya, the Contigo and Miocene (Hoorn et al. 1995;DiasdeGamero1996). Uraricoera river drainages that formerly flowed northward to Andean-derived sediments reached the Amazon fan between the Proto-Berbice river drainage were captured by the Branco 11.8 and 11.3 Mya, evidencing the breaching of the Purus arch river drainage, drastically changing the chemical composition and consequently the establishment of an east-flowing trans- of sediment inflow into the Negro River (Ferreira et al. 2007; continental Amazon River (Figueiredo 2009, 2010; Lujan and Armbruster 2011; Lujan et al. 2014). The influence Wesselingh and Hoorn 2011). Around 10–8 Mya, Andean of the Branco River sediment discharges over the distribution uplift in northern South America resulted in the rise of the of fishes in the Negro river drainage probably contributed to Vaupés arch, leading to the capture of the Negro river drainage Fluviphylax diversification. Schneider et al. (2012)proposed from the Orinoco River basin to the Amazon basin (Dias de that the Branco River acted as an ecological barrier between Gamero 1996; Winemiller and Willis 2011). The Vaupés arch middle and lower Negro River populations of the characiform represents the main divide between the Negro and the Orinoco Carnegiella strigata. According to our dated analysis, around river drainages and is considered to be responsible for some 9Mya,Fluviphylax species occurring in the Negro river drain- differentiation between the fish fauna of both river drainages age, split in two distinct groups (Fig. 3). One of these groups (Winemiller and Willis 2011). The split between Fluviphylax indicates a split of F. zonatus, from the Negro River drainage sp. A, endemic to Orinoco River basin and an adjacent area of below Branco River mouth, and Fluviphylax sp. B, from the the upper Negro river drainage (Fig. 1), and the clade contain- middle Negro drainage just above Branco River mouth ing all other species of the Amazon-Orinoco Fluviphylax (Fig. 1) at about 1 Mya, which is here assumed to be a result clade, estimated to have an age of about 10 Mya, is thus of the continuous sediment inflow of the Branco River waters contemporary with the Vaupés arch uplift. The present com- into the Negro River acting as a barrier to gene flow. mon occurrence of Fluviphylax sp. A in both Orinoco and 352 Bragança P.H.N., Costa W.J.E.M.

Acknowledgments We are grateful to P. F. Amorim, E. Henschel, and Drummond, A. J., Suchard, M. A., Xie, D., & Rambaut, A. (2012). F.P. Ottoni for the valuable help in several collecting expeditions, and to Bayesian phylogenetics with BEAUti and the BEAST 1.7. C. Gama and H. Lazzaroto for collecting additional material. We are also Molecular Biology and Evolution, 29(8), 1969–1973. grateful to H. A. Britski, M. de Pinna, N. A. Menezes, O. Oyakawa, P. Espurt, N., Baby, P., Brusset, S., Roddaz, M., Hermoza, W., Regard, V., Pruvost, S. O. Kullander J. Snoeks, E. Vreven, and Z. Gabsi for the loan Antoine, P. -O., Salas-Gismondi, R., & Bolaños, R. (2007). How of material or/and hospitality during visits to their institutions; and to P. F. does the Nazca Ridge subduction influence the modern Amazonian Amorim, A. Katz, and J.L.O. Mattos for their help with the image prep- foreland basin? Geology, 35,515–518. arations and analyses. Espurt, N., Baby, P.,Brusset, S., Roddaz, M., Hermoza, W., & Barbarand, J. (2010). The Nazca Ridge and uplift of the Fitzcarrald arch: impli- Funding information This study was supported by CNPq (Conselho cations for regionalgeology in northern South America. In C. Hoorn, Nacional de Desenvolvimento Científico e Tecnológico – Ministério de & F.P. Wesselingh (Eds.), Amazonia, Landscape and Species Ciência e Tecnologia), CAPES (Coordenação de Aperfeiçoamento de Evolution: A Look into the Past ( pp. 89–100). Blackwell: London. Pessoal de Nível Superior), and PROTAX (Programa de Capacitação Felsenstein, J. (1985). Confidence limits on phylogenies: an approach em Taxonomia). Collections were made with license number 32955–3, using the bootstrap. Evolution, 39(4), 783–791. provided by ICMBio (Instituto Chico Mendes de Conservação da Ferreira, E. J. G., Zuanon, J., Forsberg, B., Goulding, M., & Briglia- Biodiversidade). Ferreira, R. (2007). Rio Branco: Peixes, ecologia e conseração de Roraima. Amazon Conservation Association/Instituto Nacional de Pesquisas da Amazônia/Sociedade Civil Mamirauá: Manaus. Figueiredo, J., Hoorn, C., van der Ven, P., & Soares, E. (2009). Late References Miocene onset of the Amazon River and the Amazon deep-sea fan: evidence from the Foz do Amazonas basin. Geology, 37,619– 622. Arrington, D. A., & Winemiller, K. O. (2003). Diel changeover in sand- Figueiredo, J., Hoorn, C., van der Ven, P., & Soares, E. (2010). Late beach fish assemblages in a neotropical floodplain river. Journal of Miocene onset of the Amazon River and the Amazon deep-sea – Fish Biology, 63,442 459. fan: evidence from the Foz do Amazonas basin: Reply. Geology, Arrington, D. A., & Winemiller, K. O. (2006). Habitat affinity, the sea- 38, 213. sonal flood pulse, and community assembly in littoral zone of a Folmer, O., Black, M., Hoeh, W., Lutz, R., & Vrijenhoek, R. (1994). neotropical floodplain river. Journal of the North American DNA primers for amplification of mitochondrial cytochrome c ox- – Benthological Society, 25(1), 126 141. idase subunit I from diverse metazoan invertebrates. Molecular Bloom, D. D. & Lovejoy, N. R. (2011). The biogeography of marine Marine Biology and Biotechnology, 3(5), 294–299. incursions in South America. In J. S. Albert & R. E. Reis (Eds.), Gernhard, T. (2008). The conditioned reconstruction process. Journal of Historical Biogeography of Neotropical Freshwater Fishes (pp. Theoretical Biology, 253(4), 769–778. – 137 144). The Regents of the University of California. Ghedotti, M. J. (1998). Phylogeny and classification of the Anablepidae Bragança,P.H.N.,Amorim,P.F.,&Costa,W.J.E.(2018). (Teleostei: Cyprinodontiformes). In L. R. Malabarba, R. E. Reis, R. Pantanodontidae (Teleostei, Cyprinodontiformes), the sister group P. Vari, Z. M. S. Lucena, & C. A. S. Lucena (Eds.), Phylogeny and to all other cyprinodontoid killifishes as inferred by molecular data. Classification of Neotropical Fishes (pp. 561–582). Edipucrs: Porto Zoosystematics and Evolution, 94(1), 137–145. Alegre. Chenna, R., Sugawara, H., Koike, T., Lopez, R., Gibson, T. J., Higgins, Ghedotti, M. J. (2000). Phylogenetic analysis and taxonomy of the D. G., & Thompson, J. D. (2003). Multiple sequence alignment with poecilioid fishes (Teleostei: Cyprinodontiformes). Zoological the Clustal series of programs. Nucleic Acids Research., 31(13), Journal of the Linnean Society, 130,1–53. 3497–3500. Goulding, M., Leal-Carvalho, M., & Ferreira, E. (1988). Rio Negro, rich Costa, W. J. E. M. (1996). Relationships, monophyly and three new life in poor water. Amazonian diversity and foodchain ecology as species of the neotropical miniature poeciliid genus Fluviphylax seen through fish communities. the Netherlands: SPB Academic (Cyprinodontiformes: Cyprinodontoidei). Ichthyological Publishing. Exploration of Freshwaters, 7(2), 111–130. Helmstetter, A. J., Papadopulos, A. S. T., Igea, J., Van Dooren, T. J. M., Costa, W. J. E. M., & Le Bail, P. Y. (1999). Fluviphylax palikur: a new Leroi, A. M., & Savolainen, V. (2016). Viviparity stimulates diver- Poeciliid from the Rio Oiapoque basin, Northern Brazil sification in an order of fish. Nature Communications, 7, 11271. (Cyprinodontiformes: Cyprinodontoidei), with comments on mini- https://doi.org/10.1038/ncomms11271. aturization in Fluviphylax and other neotropical freshwater fishes. Ho, S. Y. W. (2014). The changing face of the molecular evolutionary Copeia, 1999,1027–1034. clock. Trends in Ecology & Evolution, 29,496–503. Costa, W. J. E. M., Amorim, P. F., & Mattos, J. L. O. (2017). Molecular Hoorn, C. (1993). Marine incursions and the influence of Andean tecton- phylogeny and timing of diversification in South American ics on the Miocene depositional history of northwestern Amazonia: Cynolebiini seasonal killifishes. Molecular Phylogenetics and results of a palynostratigraphic study. Palaeogeography, Evolution, 116,61–68. Palaeoclimatology, Palaeoecology, 105,267–309. Darriba, D., Taboada, G. L., Doallo, R., & Posada, D. (2012). jModelTest Hoorn, C. (1994). Fluvial palaeoenvironments in the intracratonic 2: more models, new heuristics and parallel computing. Nature Amazonas Basin (Early Miocene-early Middle Miocene, Methods, 9(8), 772. Colombia). Palaeogeography, Palaeoclimatology, Palaeoecology, Dias de Gamero, M. L. (1996). The changing course of the Orinoco River 109,1–54. during the Neogene: a review. Palaeogeography, Hoorn, C., Guerrero, J., Sarmiento, G. A., & Lorente, M. A. (1995). Palaeoclimatology, Palaeoecology, 123,385–402. Andean tectonics as a cause for changing drainage patterns in Donoghue, P. C. J., & Benton, M. J. (2007). Rocks and clocks: calibrating Miocene northern South America. Geology, 23,237–240. the tree of life using fossils and molecules. Trends in Ecology and Hoorn, C., Wesselingh, F. P., Hovikoski, J., & Guerrero, J. (2010). The Evolution, 22(8), 424–431. development of the Amazonian mega-wetland (Miocene; Brazil, Drummond, A. J., Ho, S. Y. W., Phillips, M. J., & Rambaut, A. (2006). Colombia, Peru, Bolivia). In C. Hoorn & F. P. Wesselingh (Eds.), Relaxed phylogenetics and dating with confidence. PLoSBiology., Amazonia, Landscape and Species Evolution: A Look into the Past 4,699–710. (pp. 123–142). Blackwell: London. Time-calibrated molecular phylogeny reveals a Miocene–Pliocene diversification in the Amazon miniature... 353

Hrbek, T., Seckinger, J., & Meyer, A. (2007). A phylogenetic and bio- Pohl, M., Milvertz, F. C., Meyer, A., & Vences, M. (2015). Multigene geographic perspective on the evolution of poeciliid fishes. phylogeny of cyprinodontiform fishes suggests continental radia- Molecular Phylogenetics and Evolution, 43,986–998. tions and a rougue taxon position of Pantanodon. Vertebrate Hubert, N., & Renno, J. F. (2006). Historical biogeography of South Zoology, 65(1), 37–44. America freshwater fishes. Journal of Biogeography, 33, 1414– Pollux, B. J. A., Meredith, R. W., Springer, M. S., Garland, T., & Reznick, 1436. D. N. (2014). The evolution of the placenta drives a shift in sexual Jaramillo, C., Romero, I., D’Apolito, C., Bayona, G., Duarte, W., selection in livebearing fish. Nature, 513,233–236. Louwye, S., Escobar, J., Luque, J., Carrillo-Briceño, J. D., Zapata, Rambaut, A., Suchard, M. A., Xie, D., & Drummond, A. J. (2013). V., Mora, A., Schouten, S., Zavada, M., Harrington, G., Ortiz, J., & Tracer,v1.5http://beast.bio.ed.ac.uk/Trace. Wesselingh, F. P. (2017). Miocene flooding events of western Reznick, D. N., Furness, A. I., Meredith, R. W., & Springer, M. S. (2017). Amazonia. Science Advances, 3(5), e1601693. https://doi.org/10. The origin and biogeographic diversification of fishes in the family 1126/sciadv.1601693. Poeciliidae. PLoSONE, 12(3), e0172546. https://doi.org/10.1371/ Lasso, C., Mojica, J. I., Usma, J. S., Maldonado-Ocampo, J. A., Do journal.pone.0172546. Nascimiento, C., Taphorn, D. C., et al. (2004). Peces de la cuenca Roberts, T. R. (1970). Description, osteology, and relationships of the del rio Orinoco. Parte L Lista de especies y distribucion por Amazonian cyprinodont fish Fluviphylax pygmaeus. Breviora, subcuencas. Biota Colombiana, 5(2), 95–158. 347,1–28. Li, C., Ortí, G., Zhang, G., & Lu, G. (2007). A practical approach to Roberts, T. R. (1972). Ecology of the fishes in the Amazon and Congo phylogenomics: the phylogeny of ray-finned fish (Actinopterygii) basins. Bulletin of the Museum of Comparative Zoolgy, 143,117– asacasestudy.BMC Evolutionary Biology, 7,44. 147. López, J. A., Chen, W. J., & Ortí, G. (2004). Esociform phylogeny. Ronquist, F., Teslenko, M., van der Mark, P., Ayres, D., Darling, A., Copeia, 3,449–564. Hohna, S., et al. (2011). MrBayes 3.2: efficient Bayesian phyloge- Lovejoy, N. R., Albert, J. S., & Crampton, W. G. R. (2006). Miocene netic inference and model choice across a large model space. marine incursions and marine/freshwater transitions: evidence from Systematic Biology, 61(3), 539–542. Neotropical fishes. Journal of South American Earth Sciences, Rosen, D. E. (1973). Suborder Cyprinodontoidei. In D.M. Cohen et al. 21(1), 5–13. (Eds.), Fishes of the western North Atlantic (pp. 229-262). Sears Lucinda,P.H.F.(2003).FamilyPoeciliidae.InR.E.Reis,S.O. Foundation for Marine Research Memoir 1, Pt. 6, Yale University. Kullander, & C. J. Ferraris Jr. (Eds.), Check List of the Freshwater Rosen, D. E., & Bailey, R. M. (1963). The poeciliid fishes Fishes of South and Central America (pp. 555–581). Edipucrs: (Cyprinodontiformes) – their structure, zoogeography and system- Porto Alegre. atics. Bulletin of the American Museum of Natural History, 126,1– Lujan, N. K. & Armbruster, J. W. (2011). The Guiana shield. In J. S. 176. Albert, & R. E. Reis (Eds.), Historical Biogeography of Schneider, C. H., Gross, M. C., Terencio, M. L., & Porto, J. I. R. (2012). Neotropical Freshwater Fishes (pp. 211–224). The Regents of the Cryptic diversity in the mtDNA of the ornamental fish Carnegiella University of California. strigata. Journal of Fish Biology, 81(4), 1210–1224. Lujan, N. K., Armbruster, J. W., Lovejoy, N. R., & Lopéz-Fernández, H. Swofford, D. L. (2003). PA UP* – Phylogenetic Analysis Using Parsi- (2014). Multilocus molecular phylogeny of the suckermouth ar- mony (*and other methods) version 4b10. Sunderland, mored catfishes (Siluriformes: Loricariidae) with a focus on subfam- Massachusetts: Sinauer Associates. ily Hypostominae. Molecular Phylogenetics and Evolution, 82, Tamura, K., Stecher, G., Peterson, D., Filipski, A., & Kumar, S. (2013). 269–288. MEGA6: molecular evolutionary genetics analysis version 6.0. Lundberg, J. G., Marshall, L. G., Guerrero, J., Horton., Malabarba, M. C. Molecular Biology and Evolution, 30(12), 2725–2729. S. L., & Wesselingh, F. (1998). The stage for neotropical fish diver- Weitzman, S. H., & Vari, R. P. (1988). Miniaturization in South America sification: a history of tropical South American rivers. In L. R. freshwater fishes: an overview and discussion. Proceedings of the Malabarba.,R.E.Reis.,RP,Vari,Z.M,Lucena,&C.A.S, Biological Society of Washington, 101,444–465. Lucena (Eds.), Phylogeny and Classification of Neotropical Fishes Wesselingh, F. P. & Hoorn, C. (2011). Geological development of the (pp. 13–48). Edipucrs: Porto Alegre. Amazon and Orinoco basins. In J.S. Albert & R.E. Reis (Eds.), Lundberg, J. G., Sabaj Pérez, M. H., Dahdul, W. M., Orangel, A., & Historical Biogeography of Neotropical Freshwater Fishes (pp. Aguilera, S. (2010). The Amazonian Neogene fish fauna. In C. 59–67). The Regents of the University of California. Hoorn & F. P. Wesselingh (Eds.), Amazonia, Landscape and Wesselingh, F. P., & Salo, J. (2006). A Miocene perspective on the evo- Species Evolution: A Look into the Past (pp. 281–301). Blackwell: lution of the Amazoniana biota. Scripta Geologica, 133,439–445. London. Whitley, G. P. (1965). Some fish genera scrutinized. Proceedings of the Maldonado-Ocampo, J. A., Vari, R. P., & Usma, J. S. (2008). Checklist of Royal Zoological Society of New South Wales, 1964–65,25–26. the freshwater fishes of Colombia. Biota Colombiana, 9(2), 143– Winemiller, K. O. & Willis, S. C. (2011). The Vaupes arch and Casiquiare 237. Canal: barriers and passages. In J. S. Albert & R. E. Reis (Eds.), Montaña, C. G., Layman, C. A., & Taphorn, D. C. (2008). Comparison of Historical Biogeography of Neotropical Freshwater Fishes (pp. fish assamblages in two littoral habitats in a neotropical Morichal 225–242). The Regents of the University of California. stream in Venezuela. Neotropical Ichthyology, 6(4), 577–582. Winemiller, K. O., Lopéz-Fernández, H., Taphorn, D. C., Nico, L. G., & Myers, G. S., & Carvalho, A. (1955). Notes on the classification and Barbarino Duque, A. (2008). Fish assemblages of the Casiquiare names of cyprinodont fishes. Tropical Fish Magazine, 4,7. River, a corridor and zoogeographical filter for dispersal between Parenti, L. R. (1981). A phylogenetic and biogeographic analysis of the Orinoco and Amazon basins. Journal of Biogeography, 35(9), cyprinodontiform fishes (Teleostei, Atherinomorpha). Bulletin of 1551–1563. the American Museum of Natural History, 168,335–357. Zwickl, D. J. (2006). Genetic algorithm approaches for the phylogenetic Parham, J. F., Donoghue, P. C., Bell, C. J., Galway, T. D., Head, J. J., analysis of large biological sequence datasets under the maximum Holroyd, P. A., et al. (2012). Best practices for justifying fossil likelihood criterion. Ph.D.dissertation, The University of Texas at calibrations. Systematic Biology, 62,346–359. Austin.