Tempo and rates of diversification in the South American cichlid genus Apistogramma (Teleostei: Perciformes: Cichlidae)

Abstract

Evaluating biodiversity and understanding the processes involved in diversification are noticeable conservation issues in fishes subject to large, sometimes illegal, ornamental trade purposes. Here, the diversity and evolutionary history of the Neotropical dwarf cichlid genus Apistogramma from several South American countries are investigated. Mitochondrial and nuclear markers are used to infer phylogenetic relationships between 31 genetically identified species. The monophyly of Apistogramma is suggested, and Apistogramma species are distributed into four clades, corresponding to three morphological lineages. Divergence times estimated with the Yule process and an uncorrelated lognormal clock dated the Apistogramma origin to the beginning of the Eocene (≈ 50 Myr) suggesting that diversification might be related to marine incursions. Our molecular dating also suggests that the Quaternary glacial cycles coincide with the phases leading to Apistogramma speciation. These past events did not influence diversification rates in the speciose genus Apistogramma, since diversification appeared low and constant through time. Further characterization of processes involved in recent Apistogramma diversity will be necessary.

About the Authors:

Christelle Tougard

Roles Conceptualization, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Visualization, Writing – original draft, Writing – review & editing

* E-mail: [email protected]

Affiliation: Institut des Sciences de l’Evolution de Montpellier (ISEM), UMR CNRS/UM/EPHE 5554, IRD 226, CIRAD 117, Montpellier, France

ORCID http://orcid.org/0000-0002-6525-0698

Carmen R. García Dávila

Roles Data curation, Investigation, Methodology, Resources, Writing – review & editing

Affiliation: Instituto de Investigaciones de la Amazonía Peruana, Laboratorio de Biología y Genética Molecular, Iquitos, Perú

Uwe Römer

Roles Writing – review & editing

Affiliation: University of Trier, Institute of Biogeography, Department of Geo-Sciences, Trier, Germany

Fabrice Duponchelle

Roles Investigation, Writing – review & editing

Affiliation: UMR Biologie des Organismes et Ecosystèmes Aquatiques, MNHN, UPMC, CNRS-7208, IRD-207, UCBN, Paris, France

Frédérique Cerqueira

Roles Investigation, Writing – review & editing

Affiliation: Institut des Sciences de l’Evolution de Montpellier (ISEM), UMR CNRS/UM/EPHE 5554, IRD 226, CIRAD 117, Montpellier, France

Emmanuel Paradis

Roles Formal analysis, Writing – review & editing

Affiliation: Institut des Sciences de l’Evolution de Montpellier (ISEM), UMR CNRS/UM/EPHE 5554, IRD 226, CIRAD 117, Montpellier, France

Bruno Guinand

Roles Writing – review & editing

Affiliation: Institut des Sciences de l’Evolution de Montpellier (ISEM), UMR CNRS/UM/EPHE 5554, IRD 226, CIRAD 117, Montpellier, France

Carlos Angulo Chávez

Roles Investigation, Writing – review & editing

Affiliation: Instituto de Investigaciones de la Amazonía Peruana, Laboratorio de Biología y Genética Molecular, Iquitos, Perú

Vanessa Salas

Roles Investigation, Writing – review & editing

Affiliation: Universidad Nacional Federico Villareal, Facultad de Oceanografía, Pesquería, Ciencias Alimentarias y Acuicultura, Lima, Perú

Sophie Quérouil

Roles Writing – review & editing

Affiliation: Institut des Sciences de l’Evolution de Montpellier (ISEM), UMR CNRS/UM/EPHE 5554, IRD 226, CIRAD 117, Montpellier, France

Susana Sirvas

Roles Writing – review & editing

Affiliation: Universidad Nacional Federico Villareal, Facultad de Oceanografía, Pesquería, Ciencias Alimentarias y Acuicultura, Lima, Perú

Jean-François Renno

Roles Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Writing – review & editing

Affiliation: UMR Biologie des Organismes et Ecosystèmes Aquatiques, MNHN, UPMC, CNRS-7208, IRD-207, UCBN, Paris, France

Introduction

The Amazon drainage system is the aquatic continental ecosystem hosting the highest fish species richness, with 2,500 species already described and some 1,000 yet to be described [1–2]. Approximately two-thirds of the Neotropical freshwater ichthyofauna occur in the Amazon drainage system [1–2]. Human activities have impacted the Amazon biodiversity since at least the pre-Columbian times and this impact has dramatically increased since the 1950s [3]. Local populations depend mainly on freshwater fishes for protein supply, and the ornamental fish trade has also contributed to the decline of some freshwater fish species in the Brazilian and Peruvian Amazon [1, 4–5]. More than anything, large-scale destruction of natural habitats (extensive road–building to allow timber exploitation, mining, gas and petroleum activities, reservoir construction and agro-industrial development) causes collateral damages on rivers, floodplains and wetlands, and thus put a high pressure on aquatic biodiversity [1, 6–10]. There is thus an urgent need to assess the global biodiversity, and especially the Amazon freshwater biodiversity.

Fish species with narrow geographical distributions are particularly threatened. This includes many Neotropical Cichlidae (subfamily Cichlinae), but especially most species of the genus Apistogramma Regan, 1913 (>100 species) [11]. Apistogramma are small fishes (dwarf cichlids) belonging to the tribe Geophagini, and characterized by a high sexual dimorphism in morphology and colour [12–13]. They occupy nearly the entire Neotropical region east of the Andes [12]. Most Apistogramma species have restricted and disconnected geographical distributions in the Amazon, Orinoco and Paraguay drainage systems of lowland tropical rainforests and open savannahs [11–12]. However, a few species, such as A. agassizii (Steindachner, 1875), A. bitaeniata Pellegrin, 1936, A. cacatuoides Hoedeman, 1951, or A. trifasciata Eigenmann and Kennedy, 1903 are rather ubiquitous, widespread and can be sympatric [11–12]. Species of Apistogramma occur in all types of water (clear, black and white waters), ranging from fast-flowing to stagnant waters [11]. They usually inhabit leaf litter on shallow banks of waters ranging from few tens (small streams) to hundreds (rivers) of km [11, 13]. Although probably many Apistogramma species still remain undescribed, molecular and morphological data suggest that this genus might be monophyletic and considered as the sister clade of Taeniacara Myers, 1935 [11, 14–18]. A cluster analysis based on coloration (notably, of lips, anterior dorsal membrane or during brood-care), as well as external morphological (such as black markings, body and fin shape, pores, dentition) and behavioural (family structure) characters established that all of the 116 Apistogramma species investigated could belong to three main groups: the steindachneri, agassizii and regani lineages [11]. A fourth lineage, including only A. diplotaenia Kullander, 1987, was suggested in a phylogenetic analysis where, however, neither the nuclear and mitochondrial markers nor the species taken into account were conveniently listed [18]. Seasonal or geological water-level fluctuations could have played an important role in Apistogramma speciation events by isolating populations and favoring the establishment of reproductive barriers [11]. A recent phylogeographic study on A. caetei Kullander, 1980 from eastern Amazonia indicated that three genetically different allopatric lineages showed a strong prezygotic isolation through female mate choice [19]. According to Ready et al. [19], the Apistogramma species richness could be seriously underestimated if future works reveal that their results are aimed to be indicative of a general trend (see also [20–21]).

Better assessment of the species diversity patterns and of the speciation processes are important conservation issues, notably for fishes subject to overfishing for ornamental trade purposes, such as Apistogramma. In this study, the diversity and evolutionary history of the Neotropical dwarf cichlid genus Apistogramma are investigated. Most attention was paid to interspecific molecular phylogenies to identify (1) the phylogenetic relationships among the species of Apistogramma and (2) putative cryptic diversity in this genus. Their evolutionary history and diversification rates are inferred from partial sequences of the mitochondrial cytochrome b and cytochrome c oxydase I genes (respectively, cytb and COI) and a nuclear marker used in several studies focused on the phylogeny of Cichlidae [22–26], the Tmo-4C4 single-copy locus (Tmo4C4).

Materials and methods

Species sampling and DNA analyses

No animal was killed specifically for the present study. Fishes were stored at the Instituto de Investigaciones de la Amazonia Peruana (IIAP) in Iquitos, Peru. Pieces of muscles and fins were taken from fishes preserved in alcohol. A permit from the Dirreccion Regional de la Produccion del Gobierno Regional de Loreto in Peru was obtained to export tissue samples to France. Documents are available upon request.

Tissue samples of Apistogramma were taken from 309 specimens for up to 41 morphologically identified species (or morphospecies) [27], depending on the analyzed dataset (S1 Table). These morphospecies were selected according to the lineages they were found to belong to in a previous cluster analysis [11]. The specimens were deposited in the Laboratorio de Biología y Genética Molecular (IIAP, Iquitos, Peru), the Museo de Historia Natural de la Universidad Nacional Mayor de San Marcos (Lima, Peru), the California Academy of Science (San Francisco, USA), the Field Museum of Natural History (Chicago, USA), and the Staatliches Museum für Tierkunde (Dresden, Germany). Tissues originated from Peru, Brazil, Venezuela, Ecuador and Bolivia. A list of the specimens with catalog numbers is provided in the S2 Table.

Total DNA was extracted from fin clips and muscle pieces preserved in 96% ethanol following standard procedures [28]. The partial Tmo4C4, cytb and COI were classically PCR-amplified (see Table 1 for primer details). Direct sequencing was carried out in both directions to confirm polymorphic sites. Sequences were produced through the technical facilities of the Platform “Genotyping and Sequencing” shared by the “Institut des Sciences de l’Evolution de Montpellier” (ISEM) and the “Centre Méditerranéen de l’Environnement et de la Biodiversité” (CeMEB) (Montpellier, France). Sequences were aligned by hand using MEGA v5.2.2 [29].

[Figure omitted. See PDF.]

Table 1. Primers used for PCR-amplification of the Tmo-4C4 nuclear locus and the cytochrome b and cytochrome c oxidase I genes.

https://doi.org/10.1371/journal.pone.0182618.t001

Phylogenetic analyses were performed on datasets including 287 original sequences of Tmo4C4 as well as 282 and 193 original sequences of cytb and COI, respectively. These datasets were completed with GenBank sequences of Satanoperca Günther, 1862, Crenicara Steindachner, 1875, Biotodoma Eigenmann and Kennedy, 1903, Gymnogeophagus Miranda-Ribeiro, 1918, Taeniacara Myers, 1935 and Geophagus Heckel, 1840, that were used as outgroup. Details about sampling sites for the original sequences, as well as the GenBank accession numbers are given in the S2 Table.

Phylogenetic analyses and species delimitation

Phylogenetic analyses were conducted on both separate (sequences) and concatenated (sequences or haplotypes) gene datasets through the technical facilities of the Platform “Montpellier Bioinformatics Biodiversity” (MBB) shared by ISEM and CeMEB. In the concatenated dataset, chimera sequences were built from Tmo4C4, cytb and COI sequences of the outgroup species. For instance, no cytb sequence was available for Crenicara punctulatum (Günther, 1863) in GenBank, only for Crenicara sp. Sequences of Tmo4C4 and COI of C. punctulatum were thus concatenated with the cytb sequence of C. sp. (S2 Table). Phylogenetic trees were reconstructed with a maximum likelihood approach (ML) using PhyML v3.0 [32] and a Bayesian inference (BI) using MrBayes v3.1.2 [33]. Best-fitting models of sequence evolution were identified for each dataset with MrModeltest v2.3 [34]. Node robustness was estimated by bootstrap percentages (BP) in ML after 1000 replicates, whereas Bayesian posterior probabilities (PP) were obtained from the 50% majority rule tree consensus after a burn-in stage of 25,000. In BI, three independent runs of five Markov chain Monte Carlo (MCMC) samplings were also performed for five million generations with trees sampled every 100 generations. Alternative hypotheses of Apistogramma lineage relationships were compared with the Shimodaira-Hasegawa Test [35] as implemented in PAUP*4.010b [36].

Species delimitation tests were performed using a multi-locus coalescent-based method implemented in BPP v3.3 [37–38]. This method takes into account incomplete lineage sorting due to ancestral polymorphism and gene vs species tree conflicts. It allows the joint estimation of Bayesian species delimitation and species tree. To validate the genetically identified species, two initial hypotheses were used: one based on a species tree reconstructed from consensus sequences obtained for each morphospecies, and the other based on the species tree (excluding the outgroup) obtained in the frame of divergence time estimates with StarBEAST2 [39] (see next section). To underline putative cryptic diversity, species delimitation tests were then performed on the species validated by BPP that included more than ten individuals. The topology of the guide tree for each species tested was extracted from the haplotype tree (Fig 1). In all cases, the analyses were based on the concatenated dataset. Several combinations of priors for ancestral population size (θs: α = 1 or 2; β = 10, 20, 100, 200, 2000) and root age (τ₀: α = 1 or 2; β = 10, 20, 100, 200, 2000) were tested. For each test, the other priors were the following: speciesdelimitation = 1, speciestree = 1, speciesmodelprior = 1, algorithm = 0, finetune ε = 2, usedata = 1, locusrate = 1, heredity = 2 (scalar values = 1 for nuclear marker and 0.25 for mitochondrial marker) and cleandata = 1. Finetune variables were automatically adjusted, and swapping rates for each parameter were checked for recommended values (0.10–0.80) [40]. Each analysis was run twice to confirm consistency between runs.

[Figure omitted. See PDF.]

Fig 1. Maximum likelihood tree reconstructed from the Apistogramma concatenated haplotype dataset of the mitochondrial cytochrome b and cytochrome c oxydase I genes and the Tmo-4C4 nuclear locus.

Haplotypes are detailed in the S2 Table. Numbers at nodes are for bootstrap percentages (≥ 50%) and posterior probabilities (≥ 0.85). Black circles indicate nodes with BP = 100% and PP = 1.00, while grey circles are for nodes with a weak support (BP < 50% and PP < 0.85). Nodes with “-”are weakly supported in maximum likelihood approach or Bayesian inference. Branches with “*” indicate short internal branches not significantly different from zero.

https://doi.org/10.1371/journal.pone.0182618.g001

Intra- and intergroup genetic divergences were estimated by the K80 distance with MEGA. As in López-Fernández et al. [17], an internal branch test was performed with MEGA on the concatenated dataset to determine whether short internal branches in the phylogeny were resolved relationships or polytomies. The neighbour-joining method was used to build a tree under the K80 model, with and without a gamma distribution (G).

Molecular dating estimates

Two scientific schools exist about the origin of the Neotropical cichlids: either related to the Gondwana tectonic fragmentation [41–45], or after trans-Atlantic dispersal from Africa [46–49]. Several attempts of molecular dating were undertaken on the sole base of the fossil record [47, 50–51]. However, the use of different calibration points, datasets, molecular markers and analytical approaches has provided different divergence time estimates for the Cichlinae and Geophagini. The oldest Neotropical cichlids known from the fossil record are from the Lumbrera Formation in Argentina [52–54]. Sedimentological, paleontological, and geochemical dating studies suggest a Middle Eocene age (47.8–41.2 Myr; Lutetian) for the uppermost section of the Lower Lumbrera Formation where a species assigned to the Geophagini, Gymnogeophagus eocenicus Malabarba et al., 2010, was discovered [53–55]. The occurrence of this fossil predates the hypothesis of a Neotropical cichlid origin after a trans-Atlantic dispersal around 29.2 Myr (34.8–25.5 Myr) [47].

The date of 44.5 ± 3.3 Myr for the occurrence of G. eocenicus in the Lower Lumbrera Formation [54] was thus used as fossil calibration point. However, from a phylogenetic standpoint, G. eocenicus has been shown to be possibly nested within the living genus Gymnogeophagus [53], and thus does not represent the most recent common ancestor of Gymnogeophagus. According to a morphometric analysis, G. eocenicus seems to be the sister species of G. rhabdotus (Hensel, 1870) and G. balzanii (Perugia, 1891) [53]. Sequences of these two latter species were available in GenBank only for cytb. In order to correctly place this calibration point for the concatenated dataset, a first analysis was thus run for a cytb sub-dataset corresponding to the specimens included in the concatenated dataset, as well as G. rhabdotus and G. balzanii. A second analysis was then run for the same cytb sub-dataset, but by removing G. rhabdotus and G. balzanii. Priors of the analyses are provided below. An estimation of the Geophagini root age by McMahan et al. [51], 51 Myr (64–40 Myr), was used as an additional calibration point. These authors used both mitochondrial and nuclear markers, as in the present study, and the oldest known fossil for several cichlid clades as calibration points rather than biogeographic hypotheses (i.e. ages related to the tectonic fragmentation of Gondwana).

To estimate the time to the most recent common ancestor (TMRCA) for the Apistogramma species, Bayesian coalescent analyses were conducted at MBB. Analyses were performed for two speciation models (Yule and birth-death) with three molecular clocks (strict, relaxed uncorrelated lognormal and relaxed uncorrelated exponential clocks) using StarBEAST2 [39]. With Tracer v.1.6 [56], speciation models and clocks were compared using the Akaike’s information criterion through MCMC (AICM) [57] to test which of them best fit our data. Normally distributed priors were used for node calibration points: the Gymnogeophagus gymnogenys (Hensel, 1870) / G. meridionalis Reis and Malabarba, 1988 node calibrated with the age of G. eocenicus (mean = 44.5, Sigma = 2); the estimation of the Geophagini root age (mean = 52, Sigma = 7.3). StarBEAST analyses were performed with five independent runs of 100 million generations with the first 10% removed as burn-in (see S1 File for details). Markov chain convergence was ascertained by visual inspection of the traces, while the stability of each run was measured using the effective sample size (ESS > 200 for all parameters) using Tracer. Results of the independent convergent runs were combined with LogCombiner v2.4.4 [58] to estimate TMRCA and 95% confidence intervals. A consensus tree was generated using TreeAnnotator v2.4.4 [58] with mean node heights as node heights option and maximum clade credibility as target tree type option.

Diversification rates

Diversification rates were estimated using BayesRate v1.63b [59] for the entire tree and for the lineages of Apistogramma as defined in the introduction from morphological and/or molecular data [11, 18]. Marginal likelihoods via the thermodynamic integration were calculated to select the best-fitting model of diversification between the pure-birth or birth-death processes, under the following parameters: 100,000 MCMC iterations per three chains for 1,000 randomly sub-sampled posterior species trees obtained with StarBEAST and excluding the outgroup. Marginal likelihoods were then compared using the AICM, as previously mentioned. Speciation (λ), extinction (μ) and diversification (λ–μ) rates through time were finally estimated with the selected model and previously mentioned parameters. The results were visualized with Tracer.

A lineages through time (LTT) plot was used to summarize the accumulation of diversity across evolutionary time for a given phylogeny. It was thus constructed with the ltt.plot function of the ape package [60–61] for R v3.3.3 [62] from the species tree (without outgroup) obtained with StarBEAST and TreeAnnotator. Predicted LTT curves (λ and μ from the BayesRate analyses) were obtained with the LTT function of ape [63], and compared to the observed LTT plot.

Results

Phylogenetic relationships and species delimitation

All new sequences were deposited in the ENA database under the accession numbers LN678825-LN678947 and LT617356-LT617520 for Tmo4C4, LN678702-LN678824 and LT617119-LT617280 for cytb, and LN678948-LN679066 and LT617281-LT617355 for COI (S2 Table).

For the separated datasets, the full alignments represented: 293 positions for Tmo4C4 with 28 phylogenetically informative sites (PIS) within the 333 Apistogramma sequences; 669 positions and 326 PIS for 315 cytb sequences; 583 positions and 265 PIS for 207 COI sequences. The concatenated dataset, which includes Tmo4C4+cytb+COI sequences of 180 individuals corresponding to 56% of the sampled specimens for, respectively, 19, 60 and 61 haplotypes, was thus 1545 nucleotides long and it has 583 PIS within the Apistogramma sequences.

The best-fitting models of nucleotide substitution were the K80 model [64] with a proportion of invariable sites (I) and a gamma distribution (G) for Tmo4C4, whereas the GTR model [65] +I+G was selected for cytb, COI and the concatenated dataset in ML. On the other hand, a mixed-model analysis (K80+I+G and GTR+I+G) was performed in BI for the concatenated dataset. Based on the subset of Apistogramma species included in the present study, the monophyly of Apistogramma was highly supported in all tree topologies, except for the one obtained in BI for the Tmo4C4 (Fig 1 and S1–S4 Figs): BP between 87% (COI) and 100% (Tmo4C4+cytb+COI); PP = 1.00. In trees reconstructed from cytb and concatenated datasets, the individuals were clustered into four monophyletic groups with weak to high support values: A1, 99% < BP < 100% and PP = 1.00; A2, less than 50% < BP < 100% and 0.81 < PP < 0.99; A3, 85% < BP < 99% and PP = 1.00; A4, 57% < BP < 94% and 0.97 < PP < 1.00. The A2, A3 and A4 clades grouped together in a trichotomic clade (88% < BP < 99% and PP = 1.00). Alternative relationships (A2+A4/A3, A3+A4/A2 or A2+A3/A4) between these three latter groups were investigated with the Shimodaira-Hasegawa Test [35]. The best ML tree differed from the tree presented in Fig 1 by placing the A2 group as the sister group of A3+A4. However, the other relationships tested (A2+A4/A3 and A2+A3/A4) were not significantly worse than the best ML tree at the 5% confidence level (0.17 < P < 0.57). In COI trees, two clades were identified: A1, BP = 99% and PP = 1.00; A2+A3+A4, BP = 84% and PP = 1.00. In Tmo4C4 trees, the phylogenetic relationships inside the Apistogramma group remained unresolved. Some individuals belonging to some undescribed morphospecies could not be attributed to a genetically identified species. They were named A. spx in all present figures and tables.

Since the genetically identified species did not fully match the morphospecies, two initial hypotheses were used in the frame of the species delimitation tests. The “morphospecies” hypothesis was not validated by the multi-locus coalescent-based method with BPP from the concatenated dataset. The BPP analyses rather suggest 26 putative species (θs: α = 2 and β = 200; τ₀: α = 2 and β = 2000; PP = 0.97) corresponding to most morphospecies except three groups. The “StarBEAST species tree” hypothesis was, on the other hand, validated (θs: α = 2 and β = 2000; τ₀: α = 2 and β = 2000; PP = 1.00). Among the valid species, six included more than ten individuals: A. agassizii, A. barlowi Römer and Hahn, 2008, A. cacatuoides, A. cinilabra Römer et al., 2011, A. eunotus Kullander, 1981, and A. sp “Morado”. For four of them (A. barlowi, A. cacatuoides, A. cinilabra, A. eunotus), the BPP analyses suggested one putative species with high posterior probabilities (θs: α = 2 and β = 20; τ₀: α = 2 and β = 200; 0.84 < PP < 1.00), while the BPP analyses suggested seven (A. sp “Morado”) or nine (A. agassizii) putative species, but with low posterior probabilities (θs: α = 2 and β = 2000; τ₀: α = 2 and β = 2000; PP = 0.32 and 0.41, respectively).

Intragroup genetic divergence ranged from 0% to 1.4% for the 31 genetically identified species from the concatenated dataset (Fig 1 and S1 Fig), while intergroup genetic divergence ranged from 1.1% to 26.7% for these species (S3 Table). The A1 group is characterized by shorter internal branches than the A2, A3 and A4 groups. However, the internal branch test indicates that most branches are significantly different from zero with length confidence probabilities higher than 95% for most interspecific branches (Fig 1) [29, 66].

Divergence time estimates

Bayesian coalescent analyses were conducted under the GTR+I+G model from the concatenated cytb and COI dataset and the K80+I+G model for the Tmo4C4 dataset. The AICM suggested that the Yule model of speciation and the relaxed uncorrelated lognormal clock were significantly best suited to our dataset. A. sp6 was removed from the dataset because only mitochondrial markers were sequenced for one individual. Results of three of the five independent runs converged and were thus combined with LogCombiner for further analyses (Fig 2A).

[Figure omitted. See PDF.]

Fig 2. Chronogram showing the divergence time estimates (A) and lineages through time plot (B) of all the Apistogramma species taken into account in the present study.

A: values at nodes and with species names reflect the time (in Myr) to the most recent common ancestor and, in brackets, the 95% confidence intervals. Values with species names are divergence times estimated from the mitochondrial dataset. Grey circles are for nodes with posterior probabilities < 0.85. PL. and IV are for the Pliocene and the Quaternary, respectively. B: the x-axis represents the time before present in Myr, while the y-axis is the number of species (N) on a logarithmic scale. The black line is for LTT plot constructed from the species tree (A), while the blue lines are for the predicted LTT curve obtained with λ = 0.072 (dashed lines are the 95% prediction interval). The insert shows two predicted LTT curves with λ = 0.072 and N = 30 (blue), and λ = 0.103 and N = 100 (red). The two curves were standardized to be compared.

https://doi.org/10.1371/journal.pone.0182618.g002

The split between the A1 and the other groups of Apistogramma from their most recent common ancestor took place at the beginning of the Eocene (≈ 50 Myr), while the split between the A2, A3 and A4 groups seems to have occurred at the end of the Eocene (≈ 39 Myr). The four identified clades or lineages began to diversify from the beginning of the Oligocene (from ≈ 32 Myr for A2 to ≈ 20 Myr for A3 and A4). The extant species included in our dataset originated within the Pleistocene (from 2.49 to 0.16 Myr).

Diversification rates

The AICM provided a strong support for the pure-birth process in diversification estimates. The net diversification rate detected with this model was 0.072 ± 0.017 (λ = 0.072 ± 0.017 and μ = 0) event/Myr. The estimate of the diversification rate was found slightly higher (0.103 ± 0.020) when the proportion of species included in the phylogeny was set to 33%. The diversification rates for each lineage are as follows: 0.184 ± 0.045 event/Myr for A1; 0.092 ± 0.027 event/Myr for A2; 0.337 ± 0.121 event/Myr for A3. No rate of diversification was estimated for the clade A4 because BayesRate does not allow this for a clade with less than three taxa.

In agreement with the above results, the observed LTT plot (Fig 2B) exhibited a pattern of constant diversification through time. The predicted LTT curve calculated with the estimated parameters showed a good fit with the observed plot, particularly for the recent times but a bit less deeper in time, though the logarithmic scale magnified the differences. Interestingly, the predicted LTT curves calculated with the two above estimates of λ (considering missing species or not) were very similar and their prediction intervals overlapped widely (insert of Fig 2B).

Discussion

The aims of the present study were, first, to investigate the phylogeny of the genus Apistogramma, a speciose group of Neotropical cichlids suffering from ornamental trade, second to better understand the origin and tempo of diversification in this genus. An extended number of morphospecies (up to 41) were included in the present molecular analyses compared to previous published studies (between 1 and 4) [14–17, 22–25, 43, 49, 51–52]. The use of only two mitochondrial (cytb and COI) and one nuclear (Tmo4C4) markers should be seen here as a limitation of our study. Nevertheless, our phylogeny based on these markers but including several representatives of the major Apistogramma lineages and species sampled appears better resolved than previous ones.

Phylogeny of Apistogramma

The monophyly of the genus Apistogramma was already established from morphological (external characters and osteology) and/or molecular (mitochondrial cytb, ND4, 16S and nuclear RAG2, Tmo-M27, Tmo-4C4 genes) datasets focused on the Geophagini [15–17]. Considering a larger and more evenly distributed number of species, the present study also suggests monophyly of the genus. Three of the four clades identified here were found to correspond to species groups described on the basis of colour, morphological and behavioural characters as well as unspecified molecular markers [11, 18]. The fourth clade could not be identified because it was not possible to include cytb, COI or Tmo4C4 sequences of its representatives (A. diplotaenia) in our datasets. Indeed, the A1 lineage includes species of the regani lineage, while the A2 and A3 lineages cluster species of the steindachneri and agassizii lineages, respectively. Species found in the A4 group, and especially A. cacatuoides, are included in the steindachneri lineage by Römer [11], while they are included in the agassizii lineage by Miller and Schliewen [18]. In the present study, the phylogenetic relationships between the A2, A3 and A4 lineages remain unresolved. Alternative hypotheses regarding the sister group relationship of A4 with A2 and A3 do not allow to favor one hypothesis rather than another. More data are required to confirm or refute the phylogenetic position of the A4 group.

Discrepancies among studies were also found within Apistogramma lineages. For instance, A. barlowi and A. nijsseni Kullander, 1979 are considered as species of the cacatuoides complex in Römer [11], whereas they are not closely related to this complex in the present study. Likewise, A. nijsseni and A. atahualpa Römer, 1997 are included in the agassizii lineage in the phylogeny of Miller and Schliewen [18], whereas both species are clustered with species of the steindachneri lineage in both Römer [11] and the present study. On the basis of morphological data, Britzke et al. [67] included A. sp. “Papagayo” and A. sp. “Pebás” in A. ortegai Britzke et al., 2014. However, in the present study (Fig 1 and S1, S3 and S4 Figs), A. sp. “Papagayo” and A. sp. “Pebás” are not sister species. This means that one of these clades, at least, is not a sub-population or a sub-species of A. ortegai, and it might be considered as a different species if samples of A. ortegai could have been included in the present dataset. Lastly, some individuals with the eunotus morphotype are clustered with A. sp. “Morado” individuals, while other individuals with the eunotus morphotype are combined in another monophyletic group. This pattern might be the result of one of the following scenarios: (1) the fixation of A. sp. “Morado” haplotypes in some A. eunotus populations by introgressive hybridization of sympatric populations from secondary contact [68–69]; (2) an incomplete lineage sorting during past speciation events [70–71]; (3) improper taxonomic identification of specimens studied. These processes are difficult to distinguish based on phylogenetic reconstructions. A further investigation is needed with more appropriate genetic markers (e.g., microsatellites or RAD-sequencing markers).

Our sequence-based phylogenetic study of 31 Apistogramma genetic species underlined widespread genetic variation within this genus. Because the intragroup upper bound (1.4%) and the intergroup lower bound (1.1%) of the range of the estimated genetic divergences are overlapping, genetic distances seem to be an uncertain criterion for delimiting closely related species [72]. Methods such as the Automatic Barcode Gap Discovery [73] could be used for this purpose but require a large number of individuals per taxon. Hereby, a multi-locus coalescent-based method was preferred to validate the genetically identified species and to evaluate putative cryptic diversity. This method suggests the occurrence of 30 putative species in the concatenated dataset. These species were found to correspond to the genetically identified species (excluding A. sp6) (Figs 1 and 2 and S1 Fig). Among the six species with a number of sampled individuals greater than ten, no cryptic diversity was underlined in A. barlowi, A. cacatuoides, A. cinilabra or A. eunotus. Putative cryptic diversity is however weakly supported in A. agassizii and A. sp. “Morado”. Such genetic diversity might reflect large phenotypic variation. For instance, A. agassizii is characterized by phenotypic plasticity for colour, patterns and body proportions, and several attempts to create new species have been proposed [12]. This kind of species with patchy isolated populations distributed in the Amazon basin could represent likely sources of investigation on cryptic diversity, and deserve more attention through phylogeographic studies or population genetics / genomics approaches.

Evolutionary history of the genus Apistogramma

The fossil record and the molecular phylogenetic evidence suggest that most lineages of freshwater fishes currently dominating Neotropical ecosystems originated by the Late Cretaceous, and started their diversification before or during the Early Paleogene [42, 74–76]. For the origin of the Cichlinae, several time estimates were proposed: at 140–120 Myr if related to the breakup of Gondwana [41–45]; 124 Myr (146–104 Myr) [50] on the basis of the same biogeographic hypothesis and the fossil record; 82 Myr (89–74 Myr) [48] or 63 Myr (74–54 Myr) [51] only on the basis of fossil calibration points. Based on paleontological and relaxed molecular-clock estimates, Friedman et al. [47] consider this origin as much younger (34.8–25.5 Myr). This latter hypothesis implies that the origin of Neotropical cichlid fish should be posterior to the origin of the oldest Neotropical fossil cichlids (between 47.8 and 41.2 Myr for Proterocara argentina, Gymnogeophagus eocenicus, and Plesioheros chauliodus from the Lumbrera Formation) [54–55]. These diverse approaches generated various interpretations for the origin of the Geophagini (between 107 Myr and 52 Myr) [48, 50–51], the split between the genera Taeniacara and Apistogramma (between 70 Myr and 33 Myr) [48, 50] or the Apistogramma origin (between 52 Myr and 15 Myr) [48, 50]. The dates obtained here for the Geophagini and Apistogramma clades are included in their previously estimated range of divergence times.

In Amazonia, major marine regressions are known notably for the Paleocene (≈ 59–55 Myr) and the Early Eocene (≈ 43–42 Myr) [74, 77]. According to our molecular dating analyses, early stages of the Apistogramma diversification seem to have occurred between marine regressions. The ancestral Apistogramma populations thus isolated because of higher salinity levels might have initiated allopatric differentiation leading to the A2 and A4 lineages during the Oligocene. On the basis of the same hypothesis, the differentiation of the A1 and A3 lineages seem to have occurred at the beginning of the first major Neogene marine incursion (≈ 20–17 Myr) [77–79]. The present-day fluvial system started then to develop and was fully established at approximately 6.8 Myr [80]. During the latest Neogene (7–2.5 Myr), lowland aquatic habitats became better-delineated drainage systems rather than a system of more or less connected wetlands and lakes. During the Quaternary, these aquatic ecosystems were strongly affected by the glacial cycles. Especially cooler temperatures (a decrease of 8°C in the Andes and of 4–5°C in Amazonian lowland) [81–82] induced, for instance, increase/decrease of seasonal water levels, sediment supply, lake-level variations, stream capture or dispersal-based habitation [75, 83]. Moreover, glacial sea level lowering induced large lateral shifts of major rivers in Early Pleistocene and a cyclic process of vertical erosion, flooding and filling in the Amazon trunk during the Middle and Late Pleistocene [83–84]. These Quaternary glacial cycles could also be at the origin of recent events of differentiation leading to the speciation of several genetically identified Apistogramma species analyzed in the present study. Quaternary sea-level oscillations and deposition of Andean sediments could be at the origin of Amazonian várzeas (freshwater swamp or flooded forests), ria (river with typical lake features because of flow velocity reduction) [84–86], and probably oxbow lakes (meander cut-off from the main river stem) considered as favourable environments for allopatric speciation [87–90].

Diversification in the genus Apistogramma

The observed LTT plot did not evidence for substantial shifts in the Apistogramma diversification, even if the rate of diversification for the A3 lineage was found slightly higher compared to the other lineages (0.337 vs 0.184 and 0.092 event/Myr). High estimates of the diversification rates could be strongly linked to the species richness of a clade [91–93]. However, the A3 lineage is not the most species-rich clade in our dataset. The rates of diversification presented here (0.072 or 0.103) for the genus Apistogramma is close to the rates observed in the Cichlidae (0.069) [51] and the Percomorpha (0.081) [94–96], but slightly higher than the rate of the Teleostei as a whole (0.041) [95]. According to McMahan et al. [51], only the Heroini were found to potentially have elevated diversification rates relative to the other Cichlinae.

Adaptive radiation is often invoked to explain the species diversification in the Cichlidae as in the East African cichlid species characterized by ecomorphological and colour variations (e.g. [97–100]) or in the Neotropical crater lake (e.g. [101–104]) and riverine [50, 105–106] cichlids. According to López-Fernández et al. [17], short branches at the base of the Geophagini clade suggest a possible early burst of evolutionary divergence considered as a pattern of adaptive radiation [107]. Compared to the Geophagini, the Apistogramma phylogenetic tree displays few basal short internal branches compared to the terminal branches. Moreover, the statistical support of the cluster generated by these internal branches was not significant, and the relationships remain thus poorly supported. This suggests a lack of phylogenetic resolution rather than fast lineage differentiation. This is largely supported by the overlapping prediction intervals of the LTT curves with different parameters and sampling scenarios. Short internal branches observed in the A1 lineage might indicate however a possible recent and rapid speciation event, or an ongoing speciation process, as also suggested by the overlap between the intra- and intergroup genetic divergences (S3 Table). Overall, the observed tree topology and LTT plot (constant diversity through time) do not seem compatible with an early burst of species diversification as postulated for the Geophagini [17, 50]. Paleogene and/or Neogene marine incursions, the establishment of the modern Amazon drainage system during the Miocene and the Quaternary glacial cycles might be as well at the origin of several vicariant events in Apistogramma. Results from the present work are however not sufficient to deeply investigate either of these hypotheses. New phylogenetic analyses on an extended dataset, the genotyping of new marker sets or eventually other appropriate approaches (e.g., behavioural studies) should be performed to confirm or to infirm these hypotheses.

Supporting information

[Figure omitted. See PDF.]

S1 Fig. Maximum-likelihood tree reconstructed from the Apistogramma concatenated sequence dataset of the mitochondrial cytochrome b and cytochrome c oxydase I genes and the Tmo-4C4 nuclear locus.

Sequences are reported in the S2 Table. Numbers at nodes are for bootstrap percentages (≥ 50%) and posterior probabilities (≥ 0.85). Black circles indicates nodes with BP = 100% and PP = 1.00, while grey circles are for nodes with a weak support (BP < 50% and PP < 0.85). Nodes with “-”are weakly supported in maximum-likelihood approach or Bayesian inference.

https://doi.org/10.1371/journal.pone.0182618.s001

(TIF)

S2 Fig. Simplified maximum-likelihood tree reconstructed from the Apistogramma sequence dataset of the Tmo-4C4 nuclear locus.

Sequences are reported in the S2 Table. Numbers at nodes are for bootstrap percentages (≥ 50%) and posterior probabilities (≥ 0.85). Grey circles are for nodes with a weak support (BP < 50% and PP < 0.85). Grey circles are for nodes with a weak support (BP < 50% and PP < 0.85). Nodes with “-”are weakly supported in maximum-likelihood approach or Bayesian inference.

https://doi.org/10.1371/journal.pone.0182618.s002

(TIF)

S3 Fig. Simplified maximum-likelihood tree reconstructed from the Apistogramma sequence dataset of the mitochondrial cytochrome b gene.

Sequences are reported in the S2 Table. Numbers at nodes are for bootstrap percentages (≥ 50%) and posterior probabilities (≥ 0.80). Black circles indicates nodes with BP = 100% and PP = 1.00, while grey circles are for nodes with a weak support (BP < 50% and PP < 0.80). Nodes with “-”are weakly supported in maximum-likelihood approach or Bayesian inference.

https://doi.org/10.1371/journal.pone.0182618.s003

(TIF)

S4 Fig. Simplified maximum-likelihood tree reconstructed from the Apistogramma sequence dataset of the mitochondrial cytochrome c oxydase I gene.

Sequences are reported in the S2 Table. Numbers at nodes are for bootstrap percentages (≥ 50%) and posterior probabilities (≥ 0.85). Black circles indicates nodes with BP = 100% and PP = 1.00, while grey circles are for nodes with a weak support (BP < 50% and PP < 0.85). Nodes with “-”are weakly supported in maximum-likelihood approach or Bayesian inference.

https://doi.org/10.1371/journal.pone.0182618.s004

(TIF)

S1 Table. Number of morphospecies and phylogenetic clades for each marker and the concatenated dataset.

https://doi.org/10.1371/journal.pone.0182618.s005

(PDF)

S2 Table. Detailed list of the Apistogramma labels and sampling localities.

Accession numbers for sequences produced in the frame of the present study and previously submitted to GenBank are provided for the cytochrome b and cytochrome c oxydase I genes and the Tmo-4C4 nuclear locus. Haplotypes from concatenated mitochondrial and nuclear markers are also listed. The tissue provider or references for sequences from GenBank are indicated. GenBank sequences of a given genus with * were combined and used as outgroup for the concatenated analysis.

https://doi.org/10.1371/journal.pone.0182618.s006

(PDF)

S3 Table. Genetic distance within and between Apistogramma species and the outgroup.

https://doi.org/10.1371/journal.pone.0182618.s007

(PDF)

S1 File. XML file generated by BEAUti v2.4.4 in order to run the StarBEAST2 analysis with BEAST v2.4.4.

https://doi.org/10.1371/journal.pone.0182618.s008

(PDF)

Acknowledgments

Thanks are expressed to the Laboratoire Mixte International “Evolution et Domestication de l’Ichtyofaune Amazonienne” (LMI EDIA) without which a part of this work would not be possible. All people who participated to field work and/or collected fish samples are gratefully acknowledged for their implication in the data acquisition process. The Dirección Regional de la Producción del Gobierno Regional de Loreto kindly gave permission to export tissue samples of Apistogramma to the IRD-IIAP. Thanks are also expressed to the team managing the platforms “Genotyping and Sequencing” and “Montpellier Bioinformatics Biodiversity” of ISEM and CeMEB. CT would like to thank Pierre-Olivier Antoine and Jean-François Agnèse (ISEM, Montpellier, France), Emmanuel Corse (IMBE, Marseille, France) and three anonymous reviewers for helpful discussions and comments on the manuscript. ISEM n°2017-175-Sud.

Citation: Tougard C, García Dávila CR, Römer U, Duponchelle F, Cerqueira F, Paradis E, et al. (2017) Tempo and rates of diversification in the South American cichlid genus Apistogramma (Teleostei: Perciformes: Cichlidae). PLoS ONE 12(9): e0182618. https://doi.org/10.1371/journal.pone.0182618

References

1. Junk WJ, Mota Soares MG, Bayley PB. Freshwater fishes of the Amazon River basin: their biodiversity, fisheries, and habitats. Aquat Ecosyst Health. 2007; 10: 153–173.

2. Reis RE, Kullander SO, Ferraris FC Jr. Check List of the Freshwater Fishes of South and Central America. Porto Alegre: Edipucrs; 2003.

3. Chazdon RL. Second growth: the promise of tropical forest regeneration in an age of deforestation. Chicago University Press, Chicago and London; 2014.

4. Moreau M-A, Coomes OT. Aquarium fish exploitation in western Amazonia: conservation issues in Peru. Environ Conserv. 2007; 34: 12–22.

5. Römer U. Kritische Bemerkungen über Angaben zu Beständen und Bestandsentwicklungen von neotropischen Kleinfischen am Beispiel von Apistogramma mendezi und Paracheirodon. In: Sioli H, editor. Reports of the Workshop and Exhibition "Gefährdete Süßwasserfische tropischer Ökosysteme", Zoologisches Forschungsinstitut und Museum Alexander Koenig, Bonn, Germany, 9–12 March 1995.

6. Castello L, Macedo MN. Large-scale degradation of Amazonian freshwater ecosystems. Global Change Biol. 2016; 22: 990–1007.

7. Finer M, Jenkins CN, Pimm SL, Keane B, Ross C. Oil and gas projects in the western Amazon: threats to wilderness, biodiversity, and indigenous peoples. PLoS One. 2008; 3: e2932. pmid:18716679

8. Foley JA, Asner GP, Costa MH, Coe MT, DeFries R, Gibbs HK et al. Forest Degradation and Loss of Ecosystem Goods and Services in the Amazon Basin. Front Ecol Environ. 2007; 5: 25–32.

9. Reis RE. Conserving the freshwater fishes of South America. International Zoo Yearbook. 2013; 47: 65–70.

10. Winemiller KO, McIntyre PB, Castello L, Fluet-Chouinard E, Giarrizzo I, Nam S et al. Balancing hydropower and biodiversity in the Amazon, Congo and Mekong. Science. 2016; 351: 128–129.

11. Römer U. Cichlid Atlas. Volume 2. Melle: Mergus Verlag; 2006.

12. Römer U. Cichlid Atlas. Volume 1. Melle: Mergus Verlag; 2000.

13. Schindler I, Staeck W. Description of Apistogramma helkeri sp. n., a new geophagine dwarf cichlid (Teleostei: Cichlidae) from the lower río Cuao (Orinoco drainage) in Venezuela. Vertebr Zool. 2013; 63: 301–306.

14. Farias IP, Ortí G, Sampaio I, Schneider H, Meyer A. The Cytochrome b Gene as a Phylogenetic Marker: The Limits of Resolution for Analyzing Relationships Among Cichlid Fishes. J Mol Evol. 2001; 53: 89–103. pmid:11479680

15. López-Fernández H, Honeycutt RL, Stiassny MJL, Winemiller KO. Morphology, molecules, and character congruence in the phylogeny of South American geophagine cichlids (Perciformes, Labroidei). Zool Scr. 2005a; 34: 627–651.

16. López-Fernández H, Honeycutt RL, Winemiller KO. Molecular phylogeny and evidence for an adaptive radiation of geophagine cichlids from South America (Perciformes: Labroidei). Mol Phylogenet Evol. 2005b; 34: 227–244.

17. López-Fernández H, Winemiller KO, Honeycutt RL. Multilocus phylogeny and rapid radiations in Neotropical cichlid fishes (Perciformes: Cichlidae: Cichlinae). Mol Phylogenet Evol. 2010; 55: 1070–1086. pmid:20178851

18. Miller M, Schliewen U. The molecular phylogeny of the genus Apistogramma—a working hypothesis. Datz special. 2005; 12: 24–25.

19. Ready JS, Sampaio I, Schneider H, Vinson C, Dos Santos T. Colour forms of Amazonian cichlid fish represent reproductively isolated species. J Evol Biol. 2006; 19: 1139–1148. pmid:16780514

20. Engelking B, Römer U, Beisenherz W. Intraspecific colour preference in mate choice by female Apistogramma cacatuoides Hoedeman, 1951 (Teleostei: Perciformes: Cichlidae). Vertebr Zool. 2010; 60: 123–138.

21. Römer U. Beisenherz W. Environmental determination of sex in Apistogramma (Cichlidae) and two other fresh-water fishes (Teleostei). J Fish Biol. 1996; 48: 714–725.

22. Streelman JT, Zardoya R, Meyer A, Karl SA. Multilocus phylogeny of cichlid fishes (Pisces: Perciformes): Evolutionary comparison of microsatellite and single-copy nuclear loci. Mol Biol Evol. 1998; 15:798–808. pmid:10766579

23. Farias IP, Ortí G, Meyer A. Total Evidence: Molecules, Morphology, and the Phylogenetics of Cichlid Fishes. J Exp Zool Part B. 2000; 288: 76–92.

24. Sparks JS, Smith WL. Phylogeny and biogeography of cichlid fishes (Teleostei: Perciformes: Cichlidae). Cladistics. 2004; 20: 501–517.

25. Chakrabarty P. Systematics and historical biogeography of Greater Antillean Cichlidae. Mol Phylogenet Evol. 2006; 39: 619–627. pmid:16495088

26. Smith WL, Chakrabarty P, Sparks JS. Phylogeny, taxonomy, and evolution of Neotropical cichlids (Teleostei: Cichlidae: Cichlinae). Cladistics. 2008; 24: 625–641.

27. Cronquist A. Once again, what is a species? In: Knutson L, editor. Biosystematics in Agriculture. New Jersey: Alleheld Osmun, Montclair; 1978. pp. 3–20.

28. Sambrook J, Fritsch EF, Maniatis T. Molecular cloning: a laboratory manual. New York: Cold Spring Harbor Laboratory Press; 1989.

29. Tamura K, Peterson D, Peterson N, Stecher G, Nei M, Kumar S. MEGA5: Molecular Evolutionary Genetics Analysis using Maximum Likelihood, Evolutionary Distance, and Maximum Parsimony Methods. Mol Biol Evol. 2011; 28: 2731–2739. pmid:21546353

30. Lovejoy N. Reinterpreting recapitulation: systematics of needlefishes and their allies (Teleostei: Beloniformes). Evolution. 2000; 54: 1349–1362. pmid:11005301

31. Streelman JT, Alfaro M, Westneat MW, Bellwood DR, Karl SA. Evolutionary history of the parrotfishes: biogeography, ecomorphology, and comparative diversity. Evolution. 2002; 56: 961–971. pmid:12093031

32. Guindon S, Dufayard JF, Lefort V, Anisimova M, Hordijk W, Gascuel O. New algorithms and methods to estimate maximum-likelihood phylogenies: assessing the performance of PhyML 3.0. Syst Biol. 2010; 57: 307–321.

33. Ronquist F, Huelsenbeck JP. MrBayes3: Bayesian phylogenetic inference under mixed models. Bioinformatics. 2003; 19: 1572–1574. pmid:12912839

34. Nylander JAA. MrModeltest v2. Program distributed by the author. Evolutionary Biology Centre, Uppsala University; 2004.

35. Shimodaira H, Hasegawa M. Multiple comparisons of log-likelihoods with applications to phylogenetic inference. Mol Biol Evol. 1999; 16: 1114–1116.

36. Swofford DL. PAUP*. Phylogenetic Analysis Using Parsimony (*and Other Methods). Version 4.0b10. Sunderland, Massachusetts: Sinauer Associates; 2002.

37. Yang Z, Rannala B. Bayesian species delimitation using multilocus sequence data. Proc Natl Acad Sci USA. 2010; 107: 9264–9269. pmid:20439743

38. Yang Z. The BPP program for species tree estimation and species delimitation. Curr Zool. 2015; 61: 854–865.

39. Ogilvie HA, Bouckaert RR, Drummond A. StarBEAST2 brings faster species tree inference and accurate estimates of substitution rates. Mol Biol Evol. 2017; msx126, pmid:28431121

40. Yang Z. Tutorial of BP&P. Version 3.3 (November 2016). http://abacus.gene.ucl.ac.uk/software/

41. Brito PM, Meunier FJ, Leal EC. Origine et diversification de l’ichtyofaune néotropicale: une revue. Cybium. 2007; 31: 139–153.

42. Farias IP, Ortí G, Sampaio I, Schneider H, Meyer A. Mitochondrial DNA Phylogeny of the Family Cichlidae: Monophyly and Fast Molecular Evolution of the Neotropical Assemblage. J Mol Evol. 1999; 48: 703–711. pmid:10229574

43. Golonka J, Bocharova NY. Hot spot activity and the break-up of Pangea. Palaeogeogr Palaeoclim Palaeoecol. 2000; 161: 49–69.

44. Pérez-Díaz L, Eagles G, Constraining South Atlantic growth with seafloor spreading data. Tectonics. 2014; 33: 1848–1873.

45. Sparks JS, Smith WL. Freshwater fishes, dispersal ability, and nonevidence: “Gondwana life rafts” to the rescue. Syst Biol. 2005; 54: 158–165. pmid:15823966

46. Briggs JC. Fishes and Birds: Gondwana life rafts reconsidered. Syst Biol. 2003; 52: 548–553. pmid:12857645

47. Friedman M, Keck BP, Dornburg A, Eytan RI, Martin CH, Hulsey CD et al. Molecular and fossil evidence place the origin of cichlid fishes long after Gondwanan rifting. Proc R Soc Lond Ser B-Biol Sci. 2014; pmid:24048155

48. Matschiner M, Musilová Z, Barth JM, Starostová Z, Salzburger W, Steel M et al. Bayesian node dating based on probabilities of fossil sampling supports trans-Atlantic dispersal of cichlid. Syst Biol. 2017; 66: 3–22.

49. Murray AM. The fossil record and biogeography of the Cichlidae (Actinopterygii: Labroidei). Biol J Linn Soc. 2001; 74: 517–532.

50. López-Fernández H, Arbour JH, Winemiller KO, Honeycutt RL. Testing for ancient adaptive radiations in Neotropical cichlid fishes. Evolution. 2013; 67: 1321–1337. pmid:23617911

51. McMahan CD, Chakrabarty P, Sparks JS, Smith WL, Davis MP. Temporal Patterns of Diversification across Global Cichlid Biodiversity (Acanthomorpha: Cichlidae). PLoS One. 2013; 8: e71162. pmid:23990936

52. Malabarba MC, Zuleta O, Del Papa C. Proterocara argentina, a new fossil cichlid from the Lumbrera Formation, Eocene of Argentina. J Vertebr Paleontol. 2006; 26: 267–275.

53. Malabarba MC, Malabarba LR, Del Papa C. Gymnogeophagus eocenicus, n. sp. (Perciformes: Cichlidae), an Eocene Cichlid from the Lumbrera Formation in Argentina. J Vertebr Paleontol. 2010; 30: 341–350.

54. Malabarba MC, Malabarba LR, López-Fernández H. On the Eocene Cichlids from the Lumbrera Formation: Additions and Implications for the Neotropical Ichthyofauna. J Vertebr Paleontol. 2014; 34: 49–58.

55. del Papa C, Kirschbaum A, Powell J, Brod A, Hongn F, Pimentel M. Sedimentological, geochemical and paleontological insights applied to continental omission surfaces: A new approach for reconstructing an Eocene foreland basin in NW Argentina. J South Am Earth Sci. 2010; 29: 327–345.

56. Rambaut A, Suchard MA, Xie D, Drummond AJ. Tracer v1.6, http://beast.bio.ed.ac.uk/Tracer; 2014.

57. Raftery A, Newton M, Satagopan J, Krivitsky P. Estimating the integrated likelihood via posterior simulation using the harmonic mean identity. In: Bernardo JM, Bayarri MJ, Berger JO, editors. Bayesian statistics. New York: Oxford University Press; 2007. pp. 1–45.

58. Drummond AJ, Suchard MA, Xie D, Rambaut A. Bayesian Phylogenetics with BEAUti and the BEAST 1.7. Mol Biol Evol. 2012; 29: 1969–1973. pmid:22367748

59. Silvestro D, Schnitzler J, Zizka G. A Bayesian framework to estimate diversification rates and their variation through time and space. BMC Evol Biol. 2011; 11: 311. pmid:22013891

60. Paradis E, Claude J, Strimmer K. APE: Analyses of phylogenetics and evolution in R language. Bioinformatics. 2004; 20: 289–290. pmid:14734327

61. Paradis E. Tutorial of the ape package version 4.1 (February 2017).

62. R Development Core Team. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria (March 2017).

63. Paradis E. Random phylogenies and the distribution of branching times. J Theor Biol. 2015; 387: 39–45. pmid:26366932

64. Kimura M. A simple method for estimating evolutionary rates of base substitutions through comparative studies of nucleotide sequences. J Mol Evol. 1980; 16: 111–120. pmid:7463489

65. Yang Z. Estimating the pattern of nucleotide substitution. J Mol Evol. 1994; 39: 105–111. pmid:8064867

66. Nei M, Kumar S. Molecular Evolution and Phylogenetics. New York: Oxford University Press; 2001.

67. Britzke R, Oliveira C, Kullander SO. Apistogramma ortegai (Teleostei: Cichlidae), a new species of cichlid fish from the Ampiyacu River in the Peruvian Amazon basin. Zootaxa. 2014; 3869: 409–419. pmid:25283927

68. Rüber L, Meyer A, Sturmbauer C, Verheyen E. Population structure in two sympatric species of the Lake Tanganyika cichlid tribe Eretmodini: evidence for introgression. Mol Ecol. 2001; 10: 1207–1225. pmid:11380878

69. Egger B, Koblmüller S, Sturmbauer C, Sefc KM. Nuclear and mitochondrial data reveal different evolutionary processes in the Lake Tanganyika cichlid genus Tropheus. BMC Evol Biol. 2007; 7: 137. pmid:17697335

70. Takahashi K, Terai Y, Nishida M, Okada N. Phylogenetic relationships and ancient incomplete lineage sorting among cichlid fishes in Lake Tanganyika as revealed by analysis of the insertion of retroposons. Mol Biol Evol. 2001; 18: 2057–2066. pmid:11606702

71. Koblmüller S, Egger B, Sturmbauer C, Sefc KM. Rapid radiation, ancient incomplete lineage sorting and ancient hybridization in the endemic Lake Tanganyika cichlid tribe Tropheini. Mol Phylogenet Evol. 2010; 55: 318–334. pmid:19853055

72. Avise JC. Phylogeography: the History and Formation of Species. Harvard University Press, Cambridge, MA; 2000.

73. Puillandre N, Lambert A, Brouillet S, Achaz G. ABGD, Automatic Barcode Gap Discovery for primary species delimitation. Mol Ecol. 2011; 21: 1864–1877. pmid:21883587

74. López-Fernández H, Albert JS. Paleogene Radiations. In: Albert JS, Reis RE, editors. Historical biogeography of Neotropical freshwater fishes. Berkeley CA: University of California Press; 2011. pp. 105–117.

75. Lundberg JG, Marshall LG, Guerrero J, Horton B, Malabarba MCSL, Wesselingh F. The Stage for Neotropical Fish Diversification: a History of Tropical South American Rivers. In: Malabarba LR, Reis RE, Vari RP, Lucena ZM, Lucena CAS, editors. Phylogeny and Classification of Neotropical Fishes. Porto Alegre: Edipucrs; 1998. pp. 13–48.

76. Hoorn C, Wesselingh FP, Hovikoski J, Guerrero J. The development of the Amazonian mega-wetland (Miocene; Brazil, Colombia, Peru, Bolivia). In: Hoorn C, Wesselingh FP, editors. Amazonia: landscape and species evolution–A look into the past. Oxford UK: Wiley-Blackwell; 2010a. pp. 124–142.

77. Bloom DD, Lovejoy NR. The Biogeography of Marine Incursions in South America. In: Albert JS, Reis RE, editors. Historical biogeography of Neotropical freshwater fishes. Berkeley CA: University of California Press; 2011. pp. 137–144.

78. Boonstra M, Ramos MIF, Lammertsma EI, Antoine P-O, Hoorn C. Marine connections of Amazonia: evidence from foraminifera and dinoflagellate cysts (early to middle Miocene, Colombia/Peru). Palaeogeogr Palaeoclim Palaeoecol. 2014; 417: 176–194.

79. Hoorn C. Marine incursions and the influence of Andean tectonics on the Miocene depositional history of northwestern Amazonia: results of a palynostratigraphic study. Palaeogeogr Palaeoclim Palaeoecol. 1993; 105: 267–309.

80. Figueiredo J, Hoorn C, van der Ven P, Soares E. Late Miocene onset of the Amazon River and the Amazon deep-sea fan: Evidence from the Foz do Amazonas Basin. Geology. 2009; 37: 619–622.

81. Colinvaux PA, Liu K-B. The Late-Quaternary climate of the western Amazon basin. Abrupt Climatic Change. 1987; 216: 113–122.

82. van der Hammen T, Hooghiemstra H. Neogene and Quaternary history of vegetation, climate, and plant diversity in Amazonia. Quat Sci Rev. 2000; 19: 725–742.

83. Wesselingh FP, Hoorn C. Geological Development of Amazon and Orinoco Basins. In: Albert JS, Reis RE, editors. Historical biogeography of Neotropical freshwater fishes. Berkeley CA: University of California Press; 2011. pp. 59–67.

84. Irion G, Kalliola R. Long-term landscape development processes in Amazonia. In: Hoorn C, Wesselingh FP, editors. Amazonia: landscape and species evolution–A look into the past. Oxford UK: Wiley-Blackwell; 2010. pp. 185–197.

85. Irion G, Müller J, Morais JO, Keim G, de Mello JN, Junk WJ. The impact of Quaternary sea level changes on the evolution of the Amazonian lowland. Hydrol Process. 2009; 23: 3168–3172.

86. Irion G, de Mello JASN, Morais J, Piedade MTF, Junk WJ, Garming L. Development of the Amazon Valley During the Middle to Late Quaternary: Sedimentological and Climatological Observations. Ecological Studies. 2011; 210: 27–42.

87. Albert JS, Petry P, Reis RE. Major Biogeographic and Phylogenetic Patterns. In: Albert JS, Reis RE, editors. Historical biogeography of Neotropical freshwater fishes. Berkeley CA: University of California Press; 2011. Pp. 21–57.

88. Lowe-McConnell RH. Fish Communities in Tropical Freshwaters: Their Distribution, Ecology, and Abundance. London: Longman Press; 1975.

89. Penczak T, Galicka W, Glowacki Ł, Koszaliński H, Kruk A, Zięba G et al. Fish assemblage changes relative to environmental factors and time in the Warta River, Poland, and its oxbow lakes. J Fish Biol. 2004; 64: 483–501.

90. Winemiller KO, Tarim S, Shormann D, Cotner JB. Fish Assemblage Structure in Relation to Environmental Variation among Brazos River Oxbow Lakes. Trans Am Fish Soc. 2000; 129: 451–468.

91. Magallón S, Sanderson MJ. Absolute diversification rates in angiosperm clades. Evolution. 2001; 55: 1762–1780. pmid:11681732

92. Bloom DD, Fikáček M, Short AEZ. Clade age and diversification rate variation explain disparity in species richness among water scavenger beetle (Hydrophilidae) lineages. PLoS One. 2014; 9: e98430. pmid:24887453

93. Wiens JJ, Lapoint RT, Whiteman NK. Herbivory increases diversification across insect clades. Nat Commun. 2015; 6: 8370. pmid:26399434

94. Near TJ, Dornburg A, Eytan RI, Keck BP, Smith WL, Kuhn KL et al. Phylogeny and tempo of diversification in the superradiation of spiny-rayed fishes. Proc Natl Acad Sci USA. 2013; 110: 12738–12743. pmid:23858462

95. Santini F, Harmon LJ, Carnevale G, Alfaro ME. Did genome duplication drive the origin of teleosts? A comparative study of diversification in ray-finned fishes. BMC Evol Biol. 2009; 9: 194. pmid:19664233

96. Vega GC, Wiens JJ. Why are there so few fish in the sea? Proc R Soc Lond Ser B-Biol Sci. 2012; 279: 2323–2329.

97. Genner MJ, Turner GF. The mbuna cichlids of Lake Malawi: a model for rapid speciation and adaptive radiation. Fish Fish. 2005; 6: 1–34.

98. Salzburger W, Meyer A. The species flocks of East African cichlid fishes: recent advances in molecular phylogenies and population genetics. Naturwissenschaften. 2004; 91: 277–290. pmid:15241604

99. Salzburger W, Mack T, Verheyen E, Meyer A. Out of Tanganyika: Genesis, explosive speciation key-innovations and phylogeography of the haplochromine cichlid fishes. BMC Evol Biol. 2005; pmid:15723698

100. Seehausen O. African cichlid fish: a model system in adaptive radiation research. Proc R Soc Lond Ser B-Biol Sci. 2006; 273: 1987–1998.

101. Barluenga M, Meyer A. Phylogeography, colonization and population history of the Midas cichlid species complex (Amphilophus spp.) in the Nicaraguan crater lakes. Evol Biol. 2010; pmid:20977752

102. Elmer KR, Fan S, Kusche H, Spreitzer ML, Kautt AF, Franchini P et al. Parallel evolution of Nicaraguan crater lake cichlid fishes via non-parallel routes. Nat Commun. 2014; pmid:25346277

103. Geiger MF, McCrary JK, Schliewen UK. Not a simple case–A first comprehensive phylogenetic hypothesis for the Midas cichlid complex in Nicaragua (Teleostei: Cichlidae: Amphilophus). Mol Phylogenet Evol. 2010; 56: 1011–1024. pmid:20580847

104. Kusche H, Recknagel H, Elmer KR, Meyer A. Crater lake cichlids individually specialize along the benthic–limnetic axis. Ecol Evol. 2014; 4: 1127–1139. pmid:24772288

105. Arbour JH, López-Fernández H. Ecological variation in South American geophagine cichlids arose during an early burst of adaptive morphological and functional evolution. Proc R Soc Lond Ser B-Biol Sci. 2013; pmid:23740780

106. Piálek L, Říčan O, Casciotta J, Almirón A, Zrzavý J. Multilocus phylogeny of Crenicichla (Teleostei: Cichlidae), with biogeography of the C. lacustris group: Species flocks as model for sympatric speciation in rivers. Mol Phylogenet Evol. 2012; 62: 46–61. pmid:21971056

107. Gravilets S, Losos JB. Adaptive Radiation: Contrasting Theory with Data. Science. 2009; 323: 732–737. pmid:19197052