Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Comparable Ages for the Independent Origins of Electrogenesis in African and South American Weakly Electric Fishes

  • Sébastien Lavoué ,

    microceb@hotmail.com

    Affiliations Institute of Oceanography, National Taiwan University, Taipei, Taiwan, Atmosphere and Ocean Research Institute, The University of Tokyo, Chiba, Japan

  • Masaki Miya,

    Affiliation Department of Zoology, Natural History Museum & Institute, Chiba, Japan

  • Matthew E. Arnegard,

    Affiliation Human Biology Division, Fred Hutchinson Cancer Research Center, Seattle, Washington, United States of America

  • John P. Sullivan,

    Affiliation Cornell University Museum of Vertebrates, Ithaca, New York, United States of America

  • Carl D. Hopkins,

    Affiliation Department of Neurobiology and Behavior, Cornell University, Ithaca, New York, United States of America

  • Mutsumi Nishida

    Affiliation Atmosphere and Ocean Research Institute, The University of Tokyo, Chiba, Japan

Comparable Ages for the Independent Origins of Electrogenesis in African and South American Weakly Electric Fishes

  • Sébastien Lavoué, 
  • Masaki Miya, 
  • Matthew E. Arnegard, 
  • John P. Sullivan, 
  • Carl D. Hopkins, 
  • Mutsumi Nishida
PLOS
x

Abstract

One of the most remarkable examples of convergent evolution among vertebrates is illustrated by the independent origins of an active electric sense in South American and African weakly electric fishes, the Gymnotiformes and Mormyroidea, respectively. These groups independently evolved similar complex systems for object localization and communication via the generation and reception of weak electric fields. While good estimates of divergence times are critical to understanding the temporal context for the evolution and diversification of these two groups, their respective ages have been difficult to estimate due to the absence of an informative fossil record, use of strict molecular clock models in previous studies, and/or incomplete taxonomic sampling. Here, we examine the timing of the origins of the Gymnotiformes and the Mormyroidea using complete mitogenome sequences and a parametric Bayesian method for divergence time reconstruction. Under two different fossil-based calibration methods, we estimated similar ages for the independent origins of the Mormyroidea and Gymnotiformes. Our absolute estimates for the origins of these groups either slightly postdate, or just predate, the final separation of Africa and South America by continental drift. The most recent common ancestor of the Mormyroidea and Gymnotiformes was found to be a non-electrogenic basal teleost living more than 85 millions years earlier. For both electric fish lineages, we also estimated similar intervals (16–19 or 22–26 million years, depending on calibration method) between the appearance of electroreception and the origin of myogenic electric organs, providing rough upper estimates for the time periods during which these complex electric organs evolved de novo from skeletal muscle precursors. The fact that the Gymnotiformes and Mormyroidea are of similar age enhances the comparative value of the weakly electric fish system for investigating pathways to evolutionary novelty, as well as the influences of key innovations in communication on the process of species radiation.

Introduction

The Mormyroidea and Gymnotiformes, the African and South American weakly electric fishes respectively, have been the object of neuroethological research for decades (e.g. [1][10]), and are emerging as excellent comparative vertebrate systems for evolutionary neurobiology. Recent work on these two groups has investigated mechanisms by which animal communication and associated nervous system functions evolve and have feedback effects on evolutionary processes. Examples include studies of reproductive character displacement [11], the role of communication in speciation [12][15], effects of the evolution of neural structures on the process of species radiation [16][18], and genetic mechanisms underlying the origins of evolutionary novelty [19], [20]. Although these areas of investigation are informed by recent phylogenetic advances [19], [21][24], inferences about the temporal context of these processes are limited by the lack of an hypothesis for the timing of the origin and early diversification of these two groups of electrogenic teleosts. The purpose of this paper is to hypothesize just such a phylogenetic timeframe for the African and South American weakly electric fishes and to discuss its implications for understanding their evolution.

Extraordinary Convergences Around a Novel Sensory and Communication System

Electroreception, the ability to sense weak electric fields, is widely distributed in non-teleost aquatic craniates (Fig. 1). Ampullary electroreceptors, which are tuned to passively produced, low frequency electric fields, are found in lineages ranging from jawless craniates (lampreys) to several groups of “ancient fishes” such as chondrichthyans, coelacanths, and sturgeons [1], [25], [26]. This pattern suggests that electroreception is an ancient sense within the Craniata. However, because several hypotheses concerning reconstruction of the evolution of electroreception are equally parsimonious, it cannot be determined whether the most recent common ancestor of all craniates was electroreceptive (Fig. 1). Within teleost fishes–by far the largest group of vertebrates with more than 31,000 species [27]–electroreception is restricted to only two distantly related groups: the Siluriphysi sensu Fink and Fink [28] (i.e., the Gymnotiformes plus Siluriformes) and the Notopteroidei (i.e., the Mormyroidea plus Notopteridae). The most parsimonious hypothesis for this peculiar pattern is that low frequency electroreception was lost in the most recent common ancestor of the Neopterygii, only to be independently re-acquired in the Siluriphysi and the Notopteroidei (Fig. 1). Whereas the African Notopteridae and Siluriformes are only passively electroreceptive, the Gymnotiformes and the Mormyroidea secondarily and independently evolved specialized electric organs dedicated to the production of weak electric discharges, in addition to high-frequency (tuberous) electroreceptors that are tuned to these signals. Together, their electric organs and tuberous electroreceptors mediate both electrical communication and “active electrolocation” (e.g. [29]), in which objects are located in space and their electrical properties sensed via distortions in the self-generated electric field [30]. Thus, the ability to first sense low frequency, passive electric fields appears to have preceded the evolution of electrogenesis in the Gymnotiformes and Mormyroidea [31], [32].

thumbnail
Figure 1. Phylogenetic distribution of electroreception within the Craniata and its evolution according to the criterion of parsimony.

The phylogenetic backbone shown here follows Nelson [140], with modifications according to Gardiner et al. [141], Lavoué et al. [119], [142], Heimberg et al. [143], Kikugawa et al. [144], Li et al. [120], and Takezaki et al. [145]. Approximate timeline adapted from the fossil record; data on electroreception and electroreceptors taken from Bullock et al. [1], [26] and Albert and Crampton [25]. Colored branches indicate electroreceptive lineages possessing electroreceptors: as modified mucous glands (orange); of the ampullary sense organ type (deep blue); of both the tuberous sense organ type and the ampullary sense organ type found in teleosts (yellow). White branches signify non-electroreceptive lineages following secondary loss of electroreceptive capability; four (possibly five) such losses are indicated by white hash marks. The origins of different forms of electroreception are indicated by black hash marks. The electroreceptive conditions of the ancestors of the Craniata and of the clade (hagfishes, lampreys) are unresolved (indicated with grey and question marks) because there are several equi-parsimonious hypotheses concerning them. The end bud electroreceptor of the lampreys and the ampullary electroreceptor of the basal gnathostomes are anatomically very different, suggesting independent origins. The tree does not map atypical reports of electroreceptive gains in single species, which are in need of further study, such as tuberous electroreceptors in a blind catfish [146]. Recently, Czech-Damal et al. [147] discovered a novel sensory organ and possible electroreceptors associated with the hairless vibrissal crypts on the snout of the Guiana Dolphin (Sotalia guianensis), which appear to be sensitive to weak D.C. electric fields on the order of 4.6 microvolts per cm. Although their studies so far involve only one captive specimen trained to respond to the presence or absence of weak electric fields, it indeed suggests that additional research is needed on the sensory capabilities of aquatic mammals that might have independently evolved electroreception. Piranha (Catoprion mento) and platypus illustrations modified from images downloaded from Wikimedia Commons; paddlefish (Polyodon spathula) illustration modified from NOAA’s Historic Fisheries Collection Catalog of Images; other fish illustrations modified from Nelson [140]; other tetrapod illustrations taken from Léo Lavoué’s coloring book.

https://doi.org/10.1371/journal.pone.0036287.g001

All mormyroids and gymnotiforms except Electrophorus electricus are referred to as “weakly” electric fishes, because the external potentials they produce are usually imperceptible to human observers without amplification. Among teleosts, the separate ability to produce “strong” electric discharges for the purposes of prey capture or defense against predators has arisen once in the gymnotiform E. electricus (known commonly as the electric eel), not at all in the African mormyroids, and once in the African electric catfish family Malapteruridae. While E. electricus is additionally capable of active electroreception using weak electric discharges, as are all other gymnotiforms, electric catfishes are not. Hereafter, for simplification, South American weakly electric fishes refer to all gymnotiforms including E. electricus.

Arising from dissimilar, non-electrogenic teleost ancestors, mormyroid and gymnotiform fishes are phenotypically similar in a number of ways. First and foremost is the general presence of electrogenesis and electroreception. Additionally, mormyroids and gymnotiforms exhibit striking convergence in specific aspects of their body form, swimming behavior, reproductive behavior, ecology, nocturnal activity patterns, electric signals, and even the neuronal algorithms used to avoid jamming of active electrolocation and communication [9], [33][41]. Some examples of convergence in body form are illustrated in Fig. 2.

thumbnail
Figure 2. Morphological convergences between African and South American electric fishes.

Mormyroid African electric fishes (left column) are facing gymnotiform South American electric fishes (right column) with similar aspects of morphology (such as elongate bodies, extended tube-like snouts, reduced eyes, and/or small mouth sizes). Anterior portion of body shown above small image of whole body (except for Petrocephalus sullivani); electric organ discharge waveform shown for every species (each trace 5 ms in total duration with head-positivity plotted upwards). (A) Mormyrops zanclirostris, 175 mm standard length (SL), Ivindo River, Gabon, (B) Sternarchorhynchus oxyrhynchus, 220 mm total length (TL), Rio Negro, Brazil; (C) Mormyrus proboscirostris, 232 mm standard length, Ubundu, Congo River, D.R. Congo; (D) Rhamphichthys sp., 305 mm TL, Rio Negro, Brazil; (E) Mormyrops anguilloides, 195 mm SL, Yangambi, Congo River, D.R. Congo; (F) Gymnotus sp., 195 mm TL, Rio Negro, Brazil; and (G) Petrocephalus sullivani, Ogooué River, Gabon; (H) Eigenmannia sp., Apure River, Venezuela. Species A–D feed on benthic invertebrates, species E, F are piscivorous, and G, H feed on pelagic invertebrates.

https://doi.org/10.1371/journal.pone.0036287.g002

Convergent evolution may be defined as the independent evolution of similar biological traits from dissimilar ancestral states in unrelated lineages. In addition to convergent evolution, weakly electric fishes can also be characterized by parallel evolution; the latter term is often used to describe independent but similar patterns of trait divergence from the same ancestral trait, regardless of whether the parallel pattern of divergence occurred in closely or distantly related lineages.

From the perspective of evolutionary developmental biology, gymnotiforms and mormyroids exhibit extraordinary evolutionary parallelism in at least two important ways. First, both gymnotiforms and mormyroids evolved novel myogenic electric organs (EOmyo) that are developmentally derived from skeletal muscle progenitor cells (myoblasts) [42][47]. Further, at genetic and molecular levels, Zakon et al. [20] and Arnegard et al. [19] demonstrated, in both groups, that the same sodium channel α-subunit paralog (i.e. gene duplicate) was co-opted from skeletal muscle for exclusive expression in EOmyo, and that similar patterns of amino acid substitution subsequently occurred in regions of the α-subunit thought to contribute to electric signal variation. Second, both groups are characterized by the origin of high frequency electroreceptors (underlying active electrolocation and electrocommunication), which are derived from similar lateral line receptor precursors [30], [48][52]. The parallel origins of these complex traits on both the sender and receiver sides of electrical signaling offers an opportunity to investigate generalized patterns (or “rules”) underlying the origins of evolutionary novelty, a topic of great current interest [53][55]. While acknowledging that weakly electric fishes are characterized by both convergent and parallel evolution, depending on one’s perspective, we refer to these fishes hereafter as exemplifying a textbook example of convergent evolution [56][58]; we do so in order to highlight the independent origins of Mormyroids and Gymnotiforms from phylogenetically unrelated and phenotypically dissimilar teleost ancestors.

Previously, Lissmann [58] proposed that convergence upon an anguilliform body form with ribbon fin propulsion, present in all gymnotiforms and some mormyroids, might have been a way to minimize bending of the body axis during active electrolocation. He reasoned that such adaptations would have been an advantage in the early evolution of active electrolocation because they would reduce the amplitude modulations in the local electric fields that might have confounded the detection of objects in the environment. Attractive as the rigid body hypothesis was at the time, recent work suggests that it may not be that important for modern extant species, which have convergently evolved cerebellum-like neural circuitry in the hindbrain capable of learning, to cancel the amplitude modulations of the electric organ discharge caused by tail movements [59][65]. Others have suggested that ribbon-fin propulsion may provide enhanced maneuverability with reduced turbulence when electrolocating and approaching prey organisms [66].

There is even more reason to believe that other aspects of convergence in body form between mormyroids and gymnotiforms result from the selection pressures imposed by their shared electrosensory and electrocommunication systems. For example, Heiligenberg [67] suggested that the elongate body form with the electric organ located far from the head and trunk might be an adaptation for extending the effective distance of active electrolocation. Similarly, Hopkins [68] suggested that extending the length of the tail with its electric organ may have been a way of increasing the voltage of the electric organ discharge and hence the active space of electric signaling, which would have been important especially in water with reduced conductivity. Stoddard [69] proposed additional adaptations involving electrogenesis and electroreception for avoiding predation by electroreceptive species. Some of the striking adaptations seen in body form, such as the elongate snouts or “trunks” of some mormyroids and gymnotiforms (e.g. Fig. 2), may have arisen secondarily, long after the origins of active electrolocation, as a consequence of the types and habits of prey organisms most readily acquired in the novel ecological niches exploited by active electrosensory predators. Other convergent evolutionary shifts, such as reduced mouth sizes and restricted gill openings (Fig. 2), likely result from the need for reducing interfering electrical emissions from electrically active epithelial tissues. Additionally, the small eye sizes of many, but certainly not all, mormyroids and gymnotiforms may be a direct result of their active electrosensory systems superseding the “passive” sense of vision in importance (Fig. 2). The convergence of electric signals on either wave discharges or pulse discharges (e.g. see Fig. S1 of Arnegard et al. [19]) has stimulated much discussion in the literature, but there is no general consensus on the causes of these remarkable cases of signal convergence.

Prior Work Attempting to Date the Convergent Origins of Mormyroids and Gymnotiforms

Uncertainty surrounding the dates of origin of the Gymnotiformes and the Mormyroidea, and the timing of important innovations in electrolocation and electrocommunication, is a function of the poor fossil record of weakly electric fishes. The most ancient gymnotiform fossil excavated to date is †Humboldtichthys kirschbaumi from the Upper Miocene (ca. 8–10 millions of years ago [Mya]) of Bolivia [70][71]. The fossil record of the Mormyroidea is known only by some teeth of †Gymnarchus sp. recently described from the late Eocene (37 Mya) of Egypt [72] and some remains of †Hyperopisus sp. from the Plio-Pleistocene (i.e., more recent than 5.3 Mya) of Lake Edward and the Semliki River in Congo [73][75]. Hence, early hypotheses for the ages of the Gymnotiformes and Mormyroidea were based on phylogenetic and biogeographical hypotheses made for the large and relatively fossil-rich groups of teleosts to which they belong: the Ostariophysi and Osteoglossomorpha, respectively. Studies based on paleontological and biogeographical considerations have achieved little consensus on the ages of origin and diversification of these two groups of electric fishes, with some placing these events before [76][78] and others after [79], [80] the complete separation of Africa and South America, dated to the period from ca. 110 to 100 Mya [81].

The first molecular efforts aimed at estimating ages of the Gymnotiformes and Mormyroidea were based on short sequences of nucleotides (<850 bp) or amino acids (<750 positions), with sparse taxonomic sampling and the use of strict molecular clocks calibrated with rates derived from other groups of fishes and/or from fossils and geological events [82], [83] (also see Table 1). In these studies, ages of the Mormyroidea and Gymnotiformes were separately estimated using distinct sets of taxa and/or character data. Alves-Gomes [82] and Kumazawa and Nishida [83] estimated the age of the stem Mormyroidea to 61–72 Mya and to 241+/−23 Mya, respectively; the age of the Gymnotiformes was estimated to 79–117 Mya by Alves-Gomes [82]. The introduction of maximum likelihood and Bayesian methods to reconstruct the divergence times using complete mitochondrial sequences yielded revised estimates for the ages of these groups (Table 1). Nevertheless, these studies still included few electric fish species/lineages, with poor coverage of the morphological and taxonomic breadths of these groups. Peng et al. [84] proposed that the crown group Gymnotiformes is 150 My old and that the gymnotiform lineage originated 197 Mya, whereas Nakatani et al. [85] pushed back the origin of the crown and stem group Gymnotiformes to 189 Mya and 226 Mya, respectively. Inoue et al. [86] estimated the age of the crown group Mormyroidea to 142 Mya and the age of the mormyroid lineage to 162 Mya. Lavoué et al. [87] estimated the age of the crown group Mormyroidea to 85.2 Mya or 136.0 Mya, depending on the calibration method considered, and the origin of the stem group Mormyroidea to 104.2 Mya or 159.7 Mya.

thumbnail
Table 1. Previously estimated ages of the Mormyroidea and the Gymnotiformes.

https://doi.org/10.1371/journal.pone.0036287.t001

None of these recent studies has employed broad enough taxonomic sampling and a sufficiently large molecular dataset to simultaneously and robustly estimate ages of the Gymnotiformes and the Mormyroidea from a single tree. Here, we re-examine the ages of these two groups using: (1) complete mitochondrial genomes as our character set; (2) a unique and extensive taxonomic sampling including several basal teleost species and 27 species of Mormyroidea and Gymnotiformes, representing all families of weakly electric teleosts; and (3) a relaxed-clock Bayesian method that infers phylogenetic relationships and divergence times simultaneously, given constraints that are enforced on the basis of multiple fossil calibration points. We go on to discuss the significance of the resulting timeframe for investigating the origins of evolutionary novelty, as well as the influence of innovations in communication on species radiation, using the unified weakly electric fish system.

Results

Phylogenetic Relationships

Maximum likelihood (ML) analyses of the different data subsets yielded similar phylogenetic results, with topological differences only occurring for a few of the relationships (Fig. 3, S1, and S2). In all ML analyses across all three data subsets, the Gymnotiformes and the Mormyroidea are each monophyletic to the exclusion of all other teleosts (bootstrap proportion [BP]  = 100%). Moreover, these two clades of weakly electric fishes are nested within two distantly related groups of Teleostei, the Ostariophysi and the Osteoglossomorpha. Because the Osteoglossomorpha (with or without the Elopomorpha) is the most basal lineage in our tree, the most recent common ancestor of the Ostariophysi and the Osteoglossomorpha is also the most recent common ancestor of the crown-group Teleostei (Fig. 3).

thumbnail
Figure 3. Best maximum likelihood tree of the Teleostei from analysis of the mt-seq data subset “12RT,” using the software RAxML.

Branch lengths are proportional to the number of substitutions per nucleotide position (scale bar  = 0.05 substitutions). Numbers at nodes give node support in terms of bootstrap proportions. The tree is rooted with Amia calva. Light grey gradient boxes highlight the Mormyroidea (African weakly electric fishes) and the Gymnotiformes (South American weakly electric fishes). Arrowheads indicate nodes for which topological differences were found compared to trees reconstructed using the two other data subsets (“123ryRT” and “123RT,” shown in Fig. S1 and S2, respectively).

https://doi.org/10.1371/journal.pone.0036287.g003

Confirming a number of other studies [32], [80], [86][89], we found that the sister group of the Mormyroidea within the Osteoglossomorpha is the family Notopteridae (BP  = 100%). In contrast to an established morphological hypothesis [28], however, we inferred that the sister group of the Gymnotiformes within the Ostariophysi is the clade (Characiformes, Siluriformes). Although this relationship appears stable in our results, it is only moderately supported by BP (from 47% to 96%).

Higher-level relationships within the Mormyroidea also agreed with previous studies: we found Gymnarchus niloticus (i.e., the Gymnarchidae) to be the sister taxon of the Mormyridae [19], [24], [90], [91] (BP  = 100%); and we found the family Mormyridae to be divided into two lineages, the Petrocephalinae (Petrocephalus) and the Mormyrinae (all remaining genera) [24], [90], [91] (BP = 100%). Within the Mormyrinae, Myomyrus appears to be the most basal lineage followed by the genus Mormyrops. These nested relationships are fully consistent with the findings of Sullivan et al. [24], Lavoué et al. [22], and Arnegard et al. [19]. The clade (Isichthys, Brienomyrus) and the genus Mormyrus do not form a monophyletic group containing these three genera as found by Lavoué et al. [22]; instead, they are sequential sister groups of all remaining mormyrins [24]. Phylogenetic relationships among the rest of the Mormyrinae in the trees are not better supported by our results than by earlier molecular studies [19], [22], [24]. We found some of these relationships to be inconsistent from one analysis to the other in the present study, and with respect to the above-mentioned molecular studies.

Across our different phylogenetic analyses, we found topological instability among the four gymnotiform lineages represented in our taxonomically limited dataset: (Gymnotus, Electrophorus); Eigenmannia; (Gymnorhamphichthys, Brachyhypopomus); and Apteronotus. According to data subset “12RT” (Fig. 3), the clade (Gymnotus, Electrophorus) appears to be the sister group to the remaining gymnotiform taxa, successively followed by Apteronotus, Eigenmannia, and the clade (Gymnorhamphichthys, Brachyhypopomus). By contrast, according to the two other data subsets (“123ryRT” and “123RT,” Fig. S1 and S2 respectively), Apteronotus appears to be the most basal group, followed by either the clade (Gymnotus, Electrophorus) or Eigenmannia. The topology from our data subset “123ryRT” is fully congruent with a gymnotiform phylogeny recently inferred by Arnegard et al [19], which they found to be robust across two independently evolving nuclear genes. When we constrained our two other topologies to that of Arnegard et al. [19], we found that the best likelihood score of the constrained trees did not differ statistically from that of our best-unconstrained tree (based on AU test results; data not shown). Thus, the general findings from our mt-seq data are unable to reject the phylogenetic hypothesis of Arnegard et al. [19].

Importantly, nodes representing the two independent origins of electrogenesis in weakly electric teleosts, the dating of which was the main aim of our study, received extremely strong topological support under all data subsets and analyses.

Divergence Time Estimation

The two methods used to calibrate our chronogram produced different age estimates for the independent origins of weakly electric teleost fishes, as well as for deeper nodes in the tree (Fig. 4 and 5). Using reconstruction #1 (with strong maximum age constraints, Fig. 4), the age of the crown-group Notopteroidei and the age of the crown-group Mormyroidea were estimated to be 110.3 Mya (95% credibility interval [CI]  = 91.7–127.2 Mya) and 93.7 Mya (CI  = 74.3–112.9 Mya), respectively. The age of the most recent common ancestor of the clade (Gymnotiformes, (Siluriformes, Characiformes)), named the Characiphysae by Wiley and Johnson [92], was estimated to be 118.9 Mya (CI  = 107.6–130.1 Mya), and the age of the crown-group Gymnotiformes was estimated to be 100.2 Mya (CI  = 84.9–115.3 Mya). The second method of reconstruction (#2, with only soft maximum age constraints, Fig. 5) yielded uniformly older age estimates for the entire tree, with the ages of the Notopteroidei, the crown-group Mormyroidea, the Characiphysae, and the crown-group Gymnotiformes estimated to 147.5 Mya (CI  = 117.9–177.9 Mya), 124.8 Mya (CI  = 97.5–155.7 Mya), 169.1 Mya (CI  = 140.5–197.3 Mya), and 143.5 Mya (CI  = 115.8–171.8 Mya), respectively.

thumbnail
Figure 4. Phylogenetic chronogram of the Teleostei based on a Bayesian relaxed clock approach using the mt-seq data subset “12RT” under the first fossil calibration strategy.

In this approach, we used seven fossil-derived calibration constraints following lognormal distributions and ten others following uniform distributions (i.e., reconstruction #1). Amia calva is used to root the tree. Light grey gradient boxes highlight the Mormyroidea and Gymnotiformes. Horizontal timescale is in millions of years ago (Mya). Only selected epoch names are given. Abbreviations: E, early; Paleo, Paleocene; Eo, Eocene; and Oligo, Oligocene. Standardized timescale colors taken from the Commission for the Geological Map of the World. 95% age credibility intervals are shown as black and grey horizontal bars (calibration constraints on corresponding nodes), yellow horizontal bars (focal nodes of interest), and white horizontal bars (all other nodes). Daggers indicate that minimum ages were used to calibrate the nodes, and adjacent numbers in brackets refer to source fossils listed in the Materials and Methods. Numbers at nodes are the posterior probability support values (shown only when <1). Timing of the separation of Africa and South America is depicted by the three insets at the top, modified from [81]. Here, “origin of electroreception” refers to the initial origin of any kind of electroreceptive system in the broadest sense (see text for elaboration).

https://doi.org/10.1371/journal.pone.0036287.g004

thumbnail
Figure 5. Phylogenetic chronogram of the Teleostei based on a Bayesian relaxed clock approach using the mt-seq data subset “12RT” under the second fossil calibration strategy.

In this approach, we used 17 fossil-derived calibration constraints following uniform distributions (i.e., reconstruction #2). 95% age credibility intervals are shown as black horizontal bars (calibration constraints on corresponding nodes), yellow horizontal bars (focal nodes of interest), and white horizontal bars (all other nodes). Daggers indicate that minimum ages were used to calibrate the nodes, and adjacent numbers in brackets refer to source fossils listed in the Materials and Methods. Dashed lines between daggers and lower age limits of corresponding nodes (within the 95% age credibility intervals) depict putative ghost lineages in the fossil record. All other details as in Fig. 4.

https://doi.org/10.1371/journal.pone.0036287.g005

With each reconstruction method, the mean ages of the Mormyroidea and the Gymnotiformes were found to be very similar to each other (<15% difference under both calibration strategies), with largely overlapping credibility intervals (Fig. 6). The mean ages of the Notopteroidei and Characiphysae were also quite similar between methods, with the Characiphysae slightly older than the Notopteroidei (Fig. 6). Nodes defining the latter two clades also define minimum age estimates for the independent origins of electroreception among teleosts (i.e., in the form of derived ampullary electroreceptors; see Fig. 1). Strikingly, the intervals between these estimates for the early origins of teleost electroreception, in this broadest sense of any form of electroreception, and the corresponding estimates for the origins of electrogenesis are quite similar for the two lineages of weakly electric fishes: the mean estimated interval is 16.6 My for mormyroids vs. 18.7 My for gymnotiforms under reconstruction #1, and 22.7 My vs. 25.6 My under reconstruction #2). Thus, the independent origins of electrogenesis may have occurred after roughly the same intervals of time following the independent origins of teleost electroreception in the broad sense. However, these nodes define only minimum ages for the origin of electroreception and electrogenesis, respectively. Each trait actually evolved somewhere along the stems that subtend those nodes, and we cannot estimate the actual origin of these novel innovations with any greater precision.

thumbnail
Figure 6. Distributions of estimated ages for focal nodes of interest under each fossil calibration scheme.

For each plot, estimated ages were sampled every 5,000 generations from two independent BEAST runs of 1×108 generations each. Resulting age histograms are shown for the estimated times of the most recent common ancestors (tMRCAs) of the Mormyroidea and the Gymnotiformes (A) under reconstruction #1 (also see Fig. 4) and (C) under reconstruction #2 (also see Fig. 5); histograms are similarly shown for tMRCAs of the Notopteroidei and the Characiphysae under (B) reconstruction #1 and (D) reconstruction #2. The span of blue bars along the vertical axis of each plot gives the 95% credibility interval for that particular age estimate. Tails of each distribution are shown in red. All time scales in millions of years ago (Mya).

https://doi.org/10.1371/journal.pone.0036287.g006

Despite these points of uncertainty, our results provide evidence for comparable and perhaps nearly simultaneous dates for the origins of African and South American weakly electric fishes. We show that the origins of these groups occurred well after the splitting of the respective lineages from their most recent common ancestor, which we date to 185.7 Mya (CI  = 171.4–201.7 Mya) under reconstruction #1 or 284.1 Mya (CI  = 252.2–313.3 Mya) under reconstruction #2.

Discussion

Dating the Two Independent Origins of Electrogenesis in Teleosts

In the present study, we have made the first effort to simultaneously infer ages for the South American Gymnotiformes and the African Mormyroidea by analyzing a single, coherent molecular dataset that includes multiple representatives of both groups, as well as each of their possible sister groups. Using whole mitochondrial genome sequences and fossil calibration points for several nodes within teleosts, we made the surprising finding that gymnotiforms and mormyroids originated independently from non-electrogenic ancestors at about the same period of time in the Earth’s history. The close occurrences of these events in time appear robust across our two methods of fossil-based calibration, although the absolute date estimates differ somewhat between the two methods. Under reconstruction #1, crown-group gymnotiforms and mormyroids are estimated to have originated 100.2 and 93.7 Mya, respectively; under reconstruction #2, the respective estimates are 143.5 and 124.8 Mya.

Constraining the trees with seven maximum ages equal to the upper age limits of the strata from which fossils were excavated (reconstruction #1) provided conservative age estimates. Not surprisingly, we found such estimates to be roughly congruent with the general temporal framework of teleost diversification established by paleoichthyologists (e.g. [93], [94]). Much older divergence time estimates were obtained when we relaxed all maximum age constraints based on our selected fossils (reconstruction #2). Under this second method of calibration, for example, the estimated origin of the crown-group Teleostei was pushed back before the Mesozoic Era, to around 284 Mya. This estimate predates the first crown group teleost fossil by more than 130 My but is largely in agreement with several other molecular dating studies of the Teleostei with similarly soft maximum age constraints [84][86], [95], [96].

At present, we have no robust means of deciding which of our divergence time reconstructions provides the best estimates for the absolute ages of the Gymnotiformes and the Mormyroidea. However, our use of two very different calibration methods brackets a reasonable range of times for them. Further studies aiming to investigate divergence times within the teleosts will likely help to refine this range. Most importantly, regardless of the method of fossil-derived calibration (and, therefore, the reconstructed timeframe of divergence), the estimated mean dates for the origins of South American and African weakly electric fishes are very similar, with largely overlapping distributions (Fig. 6).

Electrogenesis in Teleosts: Two Late by-products of the Teleostean Whole Genome Duplication

Gene duplication is thought to be an important source of raw material for the origin of novel traits [97][100]. Recent investigation of a duplicated voltage-gated sodium channel gene (Scn4aa) in electric fishes has suggested that whole genome duplication (WGD) just prior to the radiation of teleosts [95], [101], [102] contributed to the origin of novel electrogenic systems in mormyroids and gymnotiforms [19], [20], [103]. The chronogram generated from reconstruction #1 suggests that the WGD event occurred at least 85.5 My before the two independent origins of weakly electrogenic fishes (at least 140.6 My under reconstruction #2). Thus, as discussed by Arnegard et al. [19], gene duplication appears to be capable of making important contributions to parallel origins of novelty even when gene duplicates are co-opted for innovation long after the duplication event occurs.

Significance of the Similar Timing Estimated for the Independent Evolution of Weakly Electric Fishes

Myogenic electric organs possessed by mormyroids and gymnotiforms are remarkably similar in many ways, though there are also important differences between lineages and great diversity in structure and function among species [40], [42][44], [104]. By demonstrating that the two groups of weakly electric fishes originated at nearly the same time in evolutionary history, we show that very similar amounts of time passed between the WGD at the base of the teleost radiation and the origins of novel myogenic electric organs within each of these lineages. Moreover, in each of the two independent groups of weakly electric fishes, we are peering back through roughly the same amounts of time to the origins of complex novel communication systems, as well as the effects of these key innovations [13], [19] on the subsequent radiations of sub-lineages and species. Our finding of these temporal similarities adds to many other lines of evidence, which we summarize in the Introduction, suggesting that weakly electric fishes present one of the most impressive, and potentially empirically valuable, cases of convergent evolution among vertebrates.

Perhaps as significant as the contemporaneous origins of mormyroids and gymnotiforms is our striking finding that very similar periods of time may have elapsed between the origin of passive electroreception and the appearance of the myogenic electric organ (and active electroreception) in each of the two lineages of weakly electric fishes. Even when one reconstructs the origin of a trait as occurring along a particular branch in a phylogeny (using the principle of parsimony, for example), there still remains much uncertainty as to when that trait actually arose along the identified branch. While acknowledging this uncertainty, we also note that the convergent evolution of myogenic electric organs and similar electrical communication systems in two distantly related teleost groups essentially offers a rare case of a single, albeit rough, degree of freedom left over for evaluating generalizations about evolutionary processes. In this context, the similarity of intervals between the appearance of any form of electroreception, initially associated with ampullary electroreceptors tuned to low electrical frequencies, and the subsequent origin of myogenic electric organs deserves some note. Our best estimates for these intervals fall within 16–19 My or 22–26 My under reconstructions #1 and #2, respectively. We assume that selection pressures for weak electric organs could only have arisen sometime after an electrosensory system began to evolve, as electroreception is required for the detection of reafferent signals produced by electric organs. Under this assumption, the above intervals provide rough upper estimates for the time period required for evolution to construct a weak, myogenic electric organ de novo from its skeletal muscle precursor.

Both mormyroids and gymnotiforms have diversified on their respective continents to a similar degree, each group comprising close to 200 named species [27]. Using comparative methods, investigators recently inferred the positive influences that a novel communication modality (in this case, electrical communication in mormyroids) and its component neural traits can exert on the opportunity for signal evolution by increasing the number of axes of variation in signal space [13], [16], [17]. They further showed that this may have the effect of augmenting the rate of signal divergence, and thus, the tempo of species radiation. Similar comparative studies have not yet been attempted within the Gymnotiformes, though they will likely be very enlightening. When such complimentary investigations are attempted, an established timeframe–presently indicating very similar periods over which these distantly related groups have radiated–will help to further inform comparisons between mormyroids and gymnotiforms.

Geographical Origin and Early Biogeography of the Gymnotiformes and Mormyroidea

The historical biogeography of the Osteoglossomorpha and the Ostariophysi has been extensively examined in relation to the tectonic history of the Earth and the fragmentation of the supercontinent Gondwana [76][78], [85], [105][108]. Both groups are good candidates for such biogeographic studies: they are predominantly restricted to freshwater, having an intolerance to salinity (yet, several marine fossils in both groups have been discovered); both groups also exhibit worldwide intercontinental distributions, and their respective fossil records indicate ancient origins predating the fragmentation of Gondwana. However, the absence of gymnotiform and mormyroid fossils older than 35 My has limited the inferences that can be made from the fossil record about their geographical origins and early diversification.

Tectonic map reconstructions show that the splitting of Africa and South America started at about 130 Mya. These continents were fully disconnected at about 100 Mya, by which time a complete north-south marine channel was fully established [81], [109][111]. Eustatic sea level reconstructions during the period from 130 Mya to 100 Mya indicate that the level of the Earth’s oceans was on average 100 meters higher than it is today [112], [113], which likely accentuated the separation between continents. Freshwater drainages on each continent appear to have been separated by a permanent seaway by around 110 Mya, which is congruent with the observed distribution of freshwater and marine fossils at this time [114], [115]. Our two reconstructions (i.e., estimates) for the origin of the Gymnotiformes and the Mormyroidea straddle this date of 110 Mya for the complete separation of Africa and South America. Reconstruction #1 favors a post-separation origin for both crown groups (100.2 Mya and 93.7 Mya, respectively), whereas reconstruction #2 infers the ages of the gymnotiform and mormyroid crown groups to be 143.5 Mya and 124.8 Mya, respectively, corresponding to the period before the complete separation of Africa and South America. Conclusively demonstrating the origin of these groups before the final breakup of Gondwanaland, however, will require the discovery of gymnotiform or mormyroid fossils older than 110 Mya.

Our divergence time reconstructions thus illustrate the long and independent histories of both electric fish groups on each continent: the origin and diversification of crown group Gymnotiformes and Mormyroidea began at the dawn of the independent histories of South America and Africa in the early Late Cretaceous. These analyses suggest the possibility that the absence of living or fossil gymnotiforms from Africa, and of mormyroids from South America, may simply be a consequence of their time of origin: these lineages may have never co-occurred on western Gondwana, or some portion of it, before it was functionally divided.

Sister Groups of the Gymnotiformes and the Mormyroidea: Dating the Origins of Electroreception

Fink and Fink [116], in their morphological comparative study of the Ostariophysi (i.e., Otophysi plus Gonorynchiformes), proposed that the Gymnotiformes and the Siluriformes form a monophyletic group based on 22 morphological synapomorphies (shared, derived traits). In addition, Fink and Fink [116] noted that the Siluriformes and the Gymnotiformes are also the only electroreceptive fishes within the Ostariophysi: the Characiformes, Cypriniformes, and Gonorynchiformes lack any form of electroreception. The work of Fink and Fink [116] and its revised version [28] has been a cornerstone of the systematics of these fishes: it represents the first and only comprehensive cladistic study of the Ostariophysi. Only seven of the 127 morphological characters listed in Fink and Fink [116] are homoplasious, yielding an extraordinarily well-supported phylogenetic hypothesis. Despite this, our result points toward the possibility of an alternative hypothesis, with moderate but consistent support for the Gymnotiformes being the sister group to the clade (Siluriformes, Characiformes), and not to the Siluriformes alone.

While previous molecular studies have provided strong support for the Characiphysae, the clade consisting of characiforms, gymnotiforms, and siluriforms (C, G, S), none has ever recovered the Fink and Fink hypothesis of ((G, S) C). Several have reported ((S, C) G) as we do here [85], [117][122], or ((G, C) S) [123][125], or they have been mostly inconclusive [126], [127]. Phylogenetic investigation of many more taxa from the Siluriformes and the Characiformes is needed to better understand the nature of this incongruence.

Importantly, the ((S, C) G) topology does not require a change to the hypothesis that an ampullary electroreceptor-based system originated only once in the Characiphysae, although it does requires the additional step of a loss of the ampullary electroreceptors subsequently in the characiform lineage. This hypothetical situation would be similar to the one favored for the Notopteroidei, where the Asian notopterids are thought to have lost electroreception [32].

Conclusion

The origins of the South American and African weakly electric fishes have been challenging to evaluate due to the incompleteness of the fossil record and the lack of robust molecular dating methods applied to a comprehensive molecular dataset consisting of all relevant taxa. Our study provides such molecular evidence for the phylogenetically independent, but more or less contemporaneous, origins of the Gymnotiformes and Mormyroidea, at least many tens of millions of years after their most recent common ancestor lived. Moreover, our findings suggest that a similar amount of time elapsed in each group between (1) the initial origin of any form of electroreception in these lineages and (2) the subsequent origin of electrogenesis, which was accompanied by the further evolution of more complex electroreceptive systems [16]. We also found that the Gymnotiformes and the Mormyroidea arose around the time of the final fragmentation of western Gondwana, but we cannot specify whether these events occurred before or after the complete separation of Africa and South America. Given the temporally similar, yet phylogenetically independent origins of the two electrogenic groups of teleosts, it is tempting to speculate that some shared environmental condition during the early Late Cretaceous, such as the climate, may have contributed to their contemporaneous origins. While it is not possible to rigorously evaluate such an hypothesis, placing the origin and diversification of African and South American electric fishes near the beginning of the separate histories of their respective continents adds a new perspective to this extraordinary, and scientifically valuable, example of convergent evolution in vertebrates.

Materials and Methods

Terminology

“Stem group” and “crown group” are useful terms when discussing the origin (stem-) and diversification (crown-) of an extant group of organisms. A crown group includes the most recent common ancestor of a living monophyletic group plus all of its descendants, living or extinct [128]. In our terminology here, a stem group includes the most ancient common ancestor of a living monophyletic group plus all of its living and extinct descendants. In other words, a stem group includes all the descendants of an extant lineage that arose anytime after the split with its living sister lineage.

Taxon Sampling Strategy and Mitogenome Sequencing

We designed the taxonomic sampling strategy for our study to simultaneously estimate the relative ages of the stem- and crown-group gymnotiforms and mormyroids. This required the inclusion of their putative sister groups, the Siluriformes and the Notopteridae respectively, as well as a large sampling of so-called basal teleost fishes. Within the Gymnotiformes, we selected one representative from all but one of the constituent families (Sternopygidae, Apteronotidae, Gymnotidae, Hypopomidae, and Rhamphichthyidae) and two representatives from the remaining family (Gymnotidae). Within the Mormyroidea, we included Gymnarchus niloticus (Gymnarchidae) plus 19 representatives of the Mormyridae representing each major lineage [22], [24].

Seventeen complete, or nearly complete, mtDNA sequences (hereafter “mt-seqs”) of African and South American electric fishes were newly obtained for this study. In only a few cases, a short portion of the control region was undetermined due to long thymine repeats. We combined our new data with the following previously published sequences: mt-seqs for four African electric fishes (Gnathonemus petersii, Myomyrus macrops, Petrocephalus microphthalmus, and Petrocephalus soudanensis) [87]; mt-seqs for five South American electric fishes (Eigenmannia virescens, Apteronotus albifrons, Electrophorus electricus, Gymnorhamphichthys hypostomus, and Gymnotus carapo) [85], [125]; and a selection of 43 mt-seqs for other teleost species. The mt-seq of Amia calva (Amiiformes) was used to root the tree. Table S1 provides additional information on the specimens included in our study, along with accession numbers for mt-seq data archived in the DDBJ, EMBL, and GenBank databases.

Lavoué et al. [119] give detailed descriptions of the standard laboratory protocols for the long and short PCR reactions that were used to obtain complete mt-seqs from 95% ethanol-preserved fin and muscle tissue. Complete lists of primer sequences and cycling conditions for these PCR reactions are available upon request to SL.

Sequence Quality and Alignment

Individual sequences were checked for quality and concatenated to assemble a consensus sequence for each mitogenome using the Sequencher software package v.4.8 (Gene Codes). Gene content and order of the newly determined mt-seqs are typical of those found in most other teleosts. Consensus sequences were then exported for analyses using other software.

Across the 70 species considered herein, sequences at each protein-coding gene were aligned manually with respect to the translated amino acid sequence. We excluded the heterogeneous base composition ND6 gene, short ambiguous stretches of alignment at the 5′ and 3′ ends of some protein-coding genes, and all stop codons from subsequent phylogenetic analyses. The 12 S and 16 S rRNA sequences, as well as the concatenated 22 tRNA genes, were aligned with the software Proalign v.0.5 [129] using default parameter settings. Regions with posterior probabilities ≤50% were excluded from the subsequent analyses. The aligned and conditioned data matrix included 14,447 nucleotide positions in total.

Partitioned Maximum Likelihood Phylogenetic Inference

We inferred phylogenetic trees by means of partitioned maximum likelihood (ML) using the software RAxML [130] with its graphical interface, raxmlGUI 0.9Beta3 [131]. Three different mt-seq data subsets, each prepared from the full data matrix, were employed for this purpose. The first data subset (“12RT,” 10,816 positions) included concatenated nucleotide sequences from 22 transfer RNA genes (1,599 positions) and the two ribosomal RNA genes (1,955 positions) plus the first two codon positions of 12 protein-coding genes (7,262 positions); third codon positions were excluded. The second data subset (“123ryRT,” 14,447 positions) included the same set of characters as the first data subset plus only transversions (not the highly homoplasious transitions) at the third codon positions of protein-coding genes (3,631 positions); limiting the analysis to transversions was implemented by replacing third position thymines (T) and cytosines (C) with “Y,” and replacing third position guanines (G) and adenines (A) with “R.” The third data subset (“123RT,” 14,447 positions) included all positions and types of substitution. We set up three partitions (data subset “12RT”: first and second codon positions of protein-coding genes, and all positions of non-protein-coding genes [tRNAs+rRNAs]) or four partitions (data subsets “123ryRT” and “123RT”: same three partitions as for data subset “12RT” plus the third codon positions of protein-coding genes).

According to analyses done using MEGA 5.05 [132], the GTR + Γ + I model–the general time reversible model with discrete gamma distributed rate heterogeneity with allowance for a proportion of invariant sites [133]–was found to be the “best” model of sequence evolution for each partition of our dataset. However, following the cautionary remarks of Stamakis in the RAxML user’s manual [130], as well as those of Yang [134], we did not allow for invariant sites in our model. Instead, we simplified it to the GTR + Γ model as recommended by Stamakis. We performed ML heuristic phylogenetic searches under this model, with data partitioning as described above. One hundred searches were made for each of the three analyses, and the best ML tree was found in each case by comparing final likelihoods among all 100 inferred trees. To evaluate the robustness of the internal branches of each of the best ML trees, 500 bootstrap replicates were calculated for each data subset under the GTR + Γ model. We then visually assessed congruence of the tree topologies resulting from the analyses we performed using the three different data subsets.

Bayesian Phylogenetic Inference and Divergence Time Estimation

We simultaneously inferred phylogenetic trees and divergence times with 95% credibility intervals. We did so using a partitioned Bayesian method that incorporated a relaxed molecular clock, as implemented by the software BEAST v.1.6.1 [135] and its suite, including BEAUTi v.1.6.1, LogCombiner v.1.6.1, and TreeAnnotator v.1.6.1. BEAUTi was first used to build the XML input files.

Mitochondrial substitution saturation can severely mislead divergence time estimation when not corrected (e.g. [136][138]). However, the negative effects of saturation are significantly reduced by removing the highly saturated third positions and using an appropriate system to partition coding and non-coding sequences and first and second codon positions [136], [137]. We therefore restricted our divergence time analysis to the data subset “12RT.” This data subset excludes third codon positions and partitions the remaining two positions separately from a third partition erected for non-protein-coding genes. The GTR + Γ model of sequence evolution was chosen for each of the three data partitions, and parameters were unlinked between partitions. In BEAUTi, we defined 17 taxon subsets for which we constrained their respective ages (based on fossil records) to follow either a lognormal distribution or a uniform distribution, as explained below.

For each reconstruction, two independent runs of 1×108 generations each were performed using BEAST. Each run was initiated from a user-starting chronogram tree that we previously built with BEAST and the data subset “12RT” using a simple HKY model of sequence evolution, a single partition, a strict molecular clock, and a single prior age constraint for the root of the tree set at 284 Mya [95]. In each run, trees and divergence time estimates were sampled once every 5,000 generations, and each run’s parameters were checked for convergence with the software Tracer v.1.5 [135]. We also graphically determined appropriate burn-in periods for each run (at least 10% of total run length). After discarding the burn-in, the remaining tree samples from the two runs were pooled into a combined file using Logcombiner. Finally, we used TreeAnnotator to calculate the maximum clade credibility tree, along with posterior probability support and mean divergence times and their 95% credibility intervals.

Fossil Selection and Calibration Strategy

Estimates of divergence time based on molecular evidence and a relaxed molecular clock are strongly dependent on how a phylogenetic tree is calibrated by selected fossils. We used two different strategies to calibrate our chronogram. Each method was based on different assumptions regarding the adequacy of the fossil record for estimating maximum ages of ancient teleost groups. Applying both of these methods in a single study allowed us to evaluate how absolute estimates for the ages of the Gymnotiformes and the Mormyroidea vary depending on calibration constraint methodology.

The first strategy we applied (reconstruction #1), which yielded more conservative (i.e., younger) age estimates, assumed that while the fossil records of some groups of Teleostei are informative only to provide minimum age limits for some nodes on the tree, the fossil records of other groups are sufficiently rich to also be informative about maximum ages [139]. Under this strategy, prior age distributions of fossil-calibrated nodes are mixed: either they are assumed to follow a uniform distribution, in which the minimum age limit is equal to the minimum age of the stratum from which the fossil was excavated and the maximum age limit is equal to the minimum age of the root of our tree, or a lognormal distribution, with a minimum age equal to the minimum age of the stratum and a maximum age equal to the maximum age of the stratum.

The second calibration strategy (reconstruction #2) assumes that the overall fossil record of the Teleostei is so incomplete that it provides only minimum ages for the origins of selected groups [86], [96]. In this case, prior age distributions of selected nodes follow a uniform distribution in which the minimum age limit is equal to the minimum age of the stratum from which the fossil was excavated, and the maximum age limit is equal to the minimum age of the root of our tree.

For both reconstructions, we selected 17 fossils in total that we further divided in two sets under reconstruction #1. The first set is composed of seven key fossils deemed to provide the best information about both minimum and maximum age constraints. The second set is composed of ten additional fossils that are less informative and only provide minimum age constraints.

The first six fossils of the first set belong to basal teleost groups having rich and coherent fossil records, while the seventh fossil represents a non-teleost group. Specifically, the seven fossils composing the first set are the following: (1) †Baugeichthys caeruleus (Albuliformes) provided a calibration for the time of the most recent common ancestor (tMRCA) for the clade (Anguilliformes, Albuliformes); (2) †Anaethalion spp. (Elopiformes) provided a calibration point for the crown-group Elopomorpha; (3) †Atolvorator longipectoralis (Aulopiformes) provided a calibration point for the crown-group Euteleostei; (4) †Laeliichthys australis (related to Arapaima gigas and Heterotis niloticus) constrained the age of the crown-group Osteoglossidae; (5) †Yanbiania wangqingica (Hiodontiformes) provided a calibration for the crown-group Osteoglossomorpha; (6) †Tischlingerichthys viohli (a stem-group ostariophysan) constrained the age of the crown-group Otocephala; and (7) †Brachydegma caelatum (Amiiformes) constrained the age of the stem-group Teleostei.

The remaining ten fossils, which compose the second set, are the following: (8) †Santanichthys diasii provided a minimum age for the most recent common ancestor (MRCA) of the clade (Siluriformes, Characiformes), i.e. excluding the Gymnotiformes; (9) †Rubiesichthys gregalis provided a minimum age for the MRCA of the crown group Ostariophysi; (10) †Chanos leopoldi provided a minimum age for the MRCA of the clade (Chanos, (Phractolaemus, Grasseichthys); (11) the oldest bagrids (†Eomacrones wilsoni, †Nigerium gadense, †Nigerium wurnoënse) and the oldest ictalurid (†Astephus sp.) provided a minimum age for the MRCA of the clade (Pseudobagrus, Ictalurus); (12) †Estesesox foxi from the Campanian provided a minimum age for the MRCA of the clade (Esox, Salmo); (13) †Hoplopteryx and †Trachichthyoides from the Cenomanian provided a minimum age for the MRCA of the clade (Gadus, Paralichthys); (14) a fossil †Scleropages sp. provided a minimum age for the MRCA of the clade (Scleropages, Osteoglossum); (15) †Paradercetis kipalaensis provided a minimum age for the MRCA of the clade (Arapaima, Heterotis); (16) †Elopoides provided a minimum age for the MRCA of the clade (Elops, Megalops); finally (17) †Carpathichthys polonicus provided a minimum age for the clade (Alepocephalus, Platytroctes).

Details on the origins of these fossil specimens, justifications for their use in calibrating minimum (and sometimes maximum) ages for nodes on the tree, relevant references, and the resulting minimum and maximum age calibrations themselves are provided in Text S1. In making these fossil-based calibrations, we followed ages of the geological stages determined by the international stratigraphy chart of the International Commission on Stratigraphy of 2009 (available online at http://www.stratigraphy.org/).

Supporting Information

Figure S1.

Best maximum likelihood tree of the Teleostei from analysis of the mt-seq data subset “123ryRT,” using the software RAxML. Branch lengths are proportional to the number of substitutions per nucleotide position (scale bar  = 0.04 substitutions). Numbers at nodes give node support in terms of bootstrap proportions. The tree is rooted with Amia calva. Light grey gradient boxes highlight the Mormyroidea (African weakly electric fishes) and the Gymnotiformes (South American weakly electric fishes). Arrowheads indicate nodes for which topological differences were found compared to trees reconstructed using the two other data subsets (“12RT” and “123RT,” shown in Fig. 3 and S2, respectively).

https://doi.org/10.1371/journal.pone.0036287.s001

(EPS)

Figure S2.

Best maximum likelihood tree of the Teleostei from analysis of the mt-seq data subset “123RT,” using the software RAxML. Branch lengths are proportional to the number of substitutions per nucleotide position (scale bar  = 0.2 substitutions). Numbers at nodes give node support in terms of bootstrap proportions. The tree is rooted with Amia calva. Light grey gradient boxes highlight the Mormyroidea (African weakly electric fishes) and the Gymnotiformes (South American weakly electric fishes). Arrowheads indicate nodes for which topological differences were found compared to trees reconstructed using the two other data subsets (“12RT” and “123ryRT,” shown in Fig. 3 and S1, respectively).

https://doi.org/10.1371/journal.pone.0036287.s002

(EPS)

Table S1.

List of the 70 species included in this study.

https://doi.org/10.1371/journal.pone.0036287.s003

(DOCX)

Text S1.

Fossil selection for chronogram node calibration.

https://doi.org/10.1371/journal.pone.0036287.s004

(DOCX)

Acknowledgments

Preliminary results of this study were first presented in 2007 at the Electrosensory Systems satellite meeting to the 8th International Congress of Neuroethology, Vancouver, British Columbia, Canada. We thank all participants of that meeting for their comments. Permits to visit the Odzala-Kokoua National Park (Republic of the Congo), and to collect fishes and export them for this study, were granted by la Direction de la Faune et des Aires Protégées (J. Kanwe and F.L. Bockandza-Paco), la Délégation à la Recherche Scientifique et Technologique (A. Itoua-Ngaporo), and ECOFAC-composante Congo (Philippe Mortier). Victor Mamonekene, Peter B. McIntyre, Valentin Mbossi, and Eric Kinzonzi provided invaluable assistance in the field. We thank John P. Friel at the Cornell University Museum of Vertebrates for curating our voucher specimens and Carmen García-Comas and Sylvain Lenoir at the National Taiwan University for statistical advice. The small Sternarchorhynchus photograph in Figure 2 is credited to Kyle R. Luckenbill (The Academy of Natural Sciences of Drexel University).

Author Contributions

Conceived and designed the experiments: SL MEA MM MN. Performed the experiments: SL. Analyzed the data: SL MEA JPS CDH. Contributed reagents/materials/analysis tools: MN MM MEA. Wrote the paper: SL MEA JPS CDH MM.

References

  1. 1. Bullock TH, Hopkins CD, Popper AN, Fay RR, editors. (2005) Electroreception. New York: Springer Science+Business Media, Icn.
  2. 2. Carlson BA (2009) Temporal-pattern recognition by single neurons in a sensory pathway devoted to social communication behavior. J Neurosci 29: 9417–9428.
  3. 3. Heiligenberg W (1977) Principles of electrolocation and jamming avoidance in electric fish: a neuroethological approach. Berlin and New York: Springer-Verlag. 85 p.
  4. 4. Hopkins CD, Bass AH (1981) Temporal coding of species recognition signals in an electric fish. Science 212: 85–87.
  5. 5. Kawasaki M (2009) Evolution of time-coding systems in weakly electric fishes. Zool Sci 26: 587–599.
  6. 6. Krahe R, Bastian J, Chacron MJ (2008) Temporal processing across multiple topographic maps in the electrosensory system. J Neurophysiol 100: 852–867.
  7. 7. Lissman HW, Machin KE (1958) The mechanisms of object location in Gymnarchus niloticus and similar fish. J Exp Biol 35: 451–486.
  8. 8. Markham MR, McAnelly ML, Stoddard PK, Zakon HH (2009) Circadian and social cues regulate ion channel trafficking. PLoS Biol 7: e1000203.
  9. 9. Rose GJ (2004) Insights into neural mechanisms and evolution of behaviour from electric fish. Nature Rev Neurosci 5: 943–951.
  10. 10. Zakon HH (2003) Insight into the mechanisms of neuronal processing from electric fish. Curr Opin Neurobiol 13: 744–750.
  11. 11. Crampton WGR, Lovejoy NR, Waddell JC (2011) Reproductive character displacement and signal ontogeny in a sympatric assemblage of electric fish. Evolution 65: 1650–1666.
  12. 12. Arnegard ME, Bogdanowicz SM, Hopkins CD (2005) Multiple cases of striking genetic similarity between alternate electric fish signal morphs in sympatry. Evolution 59: 324–343.
  13. 13. Arnegard ME, McIntyre PB, Harmon LJ, Zelditch ML, Crampton WGR, et al. (2010) Sexual signal evolution outpaces ecological divergence during electric fish species radiation. Amer Nat 176: 335–356.
  14. 14. Feulner PGD, Kirschbaum F, Mamonekene V, Ketmaier V, Tiedemann R (2007) Adaptive radiation in African weakly electric fish (Teleostei: Mormyridae: Campylomormyrus): a combined molecular and morphological approach. J Evol Biol 20: 403–414.
  15. 15. Sullivan JP, Lavoué S, Arnegard ME, Hopkins CD (2004) AFLPs resolve phylogeny and reveal mitochondrial introgression within a species flock of African electric fish (Mormyroidea : Teleostei). Evolution 58: 825–841.
  16. 16. Carlson BA, Arnegard ME (2011) Neural innovations and the diversification of African weakly electric fishes. Commun Integr Biol 4: 720–725.
  17. 17. Carlson BA, Hasan SM, Hollmann M, Miller DB, Harmon LJ, et al. (2011) Brain evolution triggers increased diversification of electric fishes. Science 332: 583–586.
  18. 18. Lavoué S, Arnegard ME, Sullivan JP, Hopkins CD (2008) Petrocephalus of Odzala offer insights into evolutionary patterns of signal diversification in the Mormyridae, a family of weakly electrogenic fishes from Africa. J Physiol–Paris 102: 322–339.
  19. 19. Arnegard ME, Zwickl DJ, Lu Y, Zakon HH (2010) Old gene duplication facilitates origin and diversification of a new communication system–twice. Proc Natl Acad Sci, USA 107: 22172–22177.
  20. 20. Zakon HH, Lu Y, Zwickl DJ, Hillis DM (2006) Sodium channel genes and the evolution of diversity in communication signals of electric fishes: Convergent molecular evolution. Proc Natl Acad Sci, USA 103: 3675–3680.
  21. 21. Alves-Gomes JA, Ortí G, Haygood M, Heiligenberg W, Meyer A (1995) Phylogenetic analysis of the South American electric fishes (Order Gymnotiformes) and the evolution of their electrogenic system: A synthesis based on morphology, electrophysiology, and mitochondrial sequence data. Mol Biol Evol 12: 298–318.
  22. 22. Lavoué S, Sullivan JP, Hopkins CD (2003) Phylogenetic utility of the first two introns of the S7 ribosomal protein gene in African electric fishes (Mormyroidea : Teleostei) and congruence with other molecular markers. Biol J Linnean Soc 78: 273–292.
  23. 23. Lovejoy NR, Lester K, Crampton WGR, Marques FPL, Albert JS (2010) Phylogeny, biogeography, and electric signal evolution of Neotropical knifefishes of the genus Gymnotus (Osteichthyes: Gymnotidae). Mol Phylogenet Evol 54: 278–290.
  24. 24. Sullivan JP, Lavoué S, Hopkins CD (2000) Molecular systematics of the African electric fishes (Mormyroidea: Teleostei) and a model for the evolution of their electric organs. J Exp Biol 203: 665–683.
  25. 25. Albert JS, Crampton WGR (2006) Electroreception and electrogenesis. In: Evans DH, Clairbone JB, editors. The Physiology of Fishes. 3rd ed. Boca Raton: CRC Press. pp. 429–470.
  26. 26. Bullock TH, Bodznick DA, Northcutt RG (1983) The phylogenetic distribution of electroreception: Evidence for convergent evolution of a primitive vertebrate sense modality. Brain Res Rev 6: 25–46.
  27. 27. Fricke R, Eschmeyer WN, editors. (2011) Catalog of Fishes electronic version website. 29: Available: http://research.calacademy.org/ichthyology/catalog/fishcatmain.asp. Accessed 2011 Mar.
  28. 28. Fink SV, Fink WL (1996) Interrelationships of ostariophysan fishes (Teleostei). In: Stiassny MLJ, Parenti LR, Johnson GD, editors. Interrelationships of Fishes. New York: Academic Press. pp. 209–245.
  29. 29. Arnegard ME, Carlson BA (2005) Electric organ discharge patterns during group hunting by a mormyrid fish. Proc R Soc Lond [Biol] 272: 1305–1314.
  30. 30. Hopkins CD (2010) Electroreception. In: Goldstein EB, editor. Encyclopedia of Perception. Thousands Oaks, CA: Sage Publications. pp. 384–387.
  31. 31. Alves-Gomes JA (2001) The evolution of electroreception and bioelectrogenesis in teleost fish: a phylogenetic perspective. J Fish Biol 58: 1489–1511.
  32. 32. Lavoué S, Sullivan JP (2004) Simultaneous analysis of five molecular markers provides a well-supported phylogenetic hypothesis for the living bony-tongue fishes (Osteoglossomorpha : Teleostei). Mol Phylogenet Evol 33: 171–185.
  33. 33. Bullock TH, Behrend K, Heiligenberg W (1975) Comparison of jamming avoidance responses in gymnotid and gymnarchid electric fish - case of convergent evolution of behavior and its sensory basis. J Comp Physiol 103: 97–121.
  34. 34. Carlson BA (2002) Neuroanatomy of the mormyrid electromotor control system. J Comp Neurol 454: 440–455.
  35. 35. Kawasaki M (1993) Independently evolved jamming avoidance responses employ identical computational algorithms: a behavioral study of the African electric fish, Gymnarchus niloticus. J Comp Physiol [A] 173: 9–22.
  36. 36. Kirschbaum F (1987) Reproduction and development of the weakly electric fish, Pollimyrus isidori (Mormyridae, Teleostei) in captivity. Environ Biol Fishes 20: 11–31.
  37. 37. Kirschbaum F (1995) Reproduction and development in mormyriform and gymnotiform fishes. In: Moller P, editor. Electric Fishes: History and Behavior. London: Chapman and Hall. pp. 267–301.
  38. 38. Kirschbaum F, Schugardt C (2002) Reproductive strategies and developmental aspects in mormyrid and gymnotiform fishes. J Physiol–Paris 96: 557–566.
  39. 39. Marrero C, Winemiller KO (1993) Tube-snouted gymnotiform and mormyriform fishes - convergence of a specialized foraging mode in teleosts. Environ Biol Fishes 38: 299–309.
  40. 40. Moller P (1995) Electric fishes: history and behavior; Pitcher TJ, editor. London: Chapman and Hall. 584 p.
  41. 41. Winemiller KO, Adite A (1997) Convergent evolution of weakly electric fishes from floodplain habitats in Africa and South America. Environ Biol Fishes 49: 175–186.
  42. 42. Bass AH (1986) Electric organs revisited: evolution of a vertebrate communication and orientation organ. In: Bullock TH, Heiligenberg W, editors. Electroreception. New York: John Wiley and Sons. pp. 13–70.
  43. 43. Bennett MVL (1971) Electric organs. In: Hoar WS, Randall DJ, editors. Fish Physiology. New York: Academic Press. pp. 347–491.
  44. 44. Kirschbaum F (1977) Electric-organ ontogeny: distinct larval organ precedes adult organ in weakly electric fish. Naturwissenschaften 64: 387–388.
  45. 45. Kirschbaum F, Schwassmann HO (2008) Ontogeny and evolution of electric organs in gymnotiform fish. J Physiol-Paris 102: 347–356.
  46. 46. Szabo TH (1960) Development of the electric organ of Mormyridae. Nature 188: 760–762.
  47. 47. Zakon HH, Unguez GA (1999) Development and regeneration of the electric organ. J Exp Biol 202: 1427–1434.
  48. 48. Gibbs MA (2004) Lateral line receptors: where do they come from developmentally and where is our research going? Brain Behav Evol 64: 163–181.
  49. 49. Modrell MS, Bemis WE, Northcutt RG, Davis MC, Baker CVH (2011) Electrosensory ampullary organs are derived from lateral line placodes in bony fishes. Nat Commun 2: 496.
  50. 50. Modrell MS, Buckley D, Baker CVH (2011) Molecular analysis of neurogenic placode development in a basal ray-finned fish. Genesis 49: 278–294.
  51. 51. Zakon HH (1986) The electroreceptive periphery. In: Bullock TH, Heiligenberg W, editors. Electroreception. New York: John Wiley and Sons. pp. 103–156.
  52. 52. Zakon HH (1988) The electroreceptors: diversity in structure and function. In: Atema J, Fay RR, Popper AN, Tavolga WN, editors. Sensory Biology of Aquatic Animals. New York: Springer. pp. 813–850.
  53. 53. Monteiro A, Podlaha O (2009) Wings, horns, and butterfly eyespots: how do complex traits evolve? Plos Biol 7: e1000037.
  54. 54. Shubin N, Tabin C, Carroll S (2009) Deep homology and the origins of evolutionary novelty. Nature 457: 818–823.
  55. 55. Wagner GP, Lynch VJ (2010) Evolutionary novelties. Curr Biol 20: R48–R52.
  56. 56. Hopkins CD (1995) Convergent designs for electrogenesis and electroreception. Curr Opin Neurobiol 5: 769–777.
  57. 57. Nishikawa KC (2002) Evolutionary convergence in nervous systems: insights from comparative phylogenetic studies. Brain Behav Evol 59: 240–249.
  58. 58. Lissman HW (1958) On the function and evolution of electric organs in fish. J Exp Biol 35: 156–191.
  59. 59. Bell CC (1981) An efference copy which is modified by reafferent input. Science 214: 450–453.
  60. 60. Bell CC (1982) Properties of a modifiable efference copy in an electric fish. J Neurophysiol 47: 1043–1056.
  61. 61. Bell CC (2001) Memory-based expectations in electrosensory systems. Curr Opin Neurobiol 11: 481–487.
  62. 62. Bell CC (2002) Evolution of cerebellum-like structures. Brain Behav Evol 59: 312–326.
  63. 63. Bastian J (1995) Pyramidal-cell plasticity in weakly electric fish: a mechanism for attenuating responses to reafferent electrosensory inputs. J Comp Physiol A 176: 63–78.
  64. 64. Bastian J (1996) Plasticity in an electrosensory system. I. General features of a dynamic sensory filter. J Neurophysiol 76: 2483–2496.
  65. 65. Bastian J (1996) Plasticity in an electrosensory system. II. Postsynaptic events associated with a dynamic sensory filter. J Neurophysiol 76: 2497–2507.
  66. 66. Shirgaonkar AA, Curet OM, Patankar NA, MacIver MA (2008) The hydrodynamics of ribbon-fin propulsion during impulsive motion. J Exp Biol 211: 3490–3503.
  67. 67. Heiligenberg W (1974) Electrolocation and jamming avoidance in a Hypopygus (Rhamphichthyidae, Gymnotoidei), an electric fish with pulse-type discharges. J Comp Physiol 91: 223–240.
  68. 68. Hopkins CD (1999) Design features for electric communication. J Exp Biol 202: 1217–1228.
  69. 69. Stoddard PK (1999) Predation enhances complexity in the evolution of electric fish signals. Nature 400: 254–256.
  70. 70. Gayet M, Meunier FJ (1991) Première découverte de Gymnotiformes fossiles (Pisces, Ostariophysi) dans le Miocène supérieur de Bolivie. C R Acad Sci III 313: 471–476.
  71. 71. Gayet M, Meunier FJ (2000) Rectification of the nomenclature of the genus name Ellisella Gayet & Meunier, 1991 (Teleostei, Ostariophysi, Gymnotiformes) in Humboldtichthys nom. nov. Cybium 24: 104.
  72. 72. Murray AM, Cook TD, Attia YS, Chatrath P, Simons EL (2010) A freshwater ichthyofauna from the late Eocene Birket Qarun Formation, Fayum, Egypt. J Vert Paleontol 30: 665–680.
  73. 73. Greenwood PH (1959) Quaternary fish-fossils. Exploration du Parc National Albert, Mission J de Heinzelin de Braucourt (1950). Bruxels: Institut des Parcs Nationaux du Congo Belge. pp. 1–80.
  74. 74. Stewart KM (2009) Fossil fish from the Nile River and its southern basins. In: Dumont HJ, editor. The Nile: Origin, Environments, Limnology and Human Use: Springer Science + Business Media B.V. pp. 677–704.
  75. 75. Van Couvering JAH (1977) Early records of freshwater fishes in Africa. Copeia 1977: 163–166. pp. 163–166.
  76. 76. Briggs JC (1979) Ostariophysan zoogeography - alternative hypothesis. Copeia 1979: 111–118. pp. 111–118.
  77. 77. Lundberg JG (1993) African-South American freshwater fish clades and continental drift: problems with a paradigm. In: Goldblatt P, editor. Biological Relationships between Africa and South America. New Haven, CT, USA: Yale University Press. pp. 156–199.
  78. 78. Patterson C (1975) The distribution of Mesozoic freshwater fishes. Mem Mus Natl Hist Nat (Paris) 88: 155–174.
  79. 79. Gosline WA (1975) A reexamination of the similarities between the freshwater fishes of Africa and South America. Mem Mus Natl Hist Nat (Paris) 88: 146–154.
  80. 80. Li G-Q, Wilson MVH (1996) Phylogeny of Osteoglossomorpha. In: Stiassny MLJ, Parenti LR, Johnson GD, editors. Interrelationships of fishes. New York: Academic press. pp. 163–174.
  81. 81. Lawver LA, Dalziel IWD, Gahagan LM (2007) Plates 2006 - atlas of plate reconstructions (750Ma to present day). Austin: University of Texas, Institute for Geophysics.
  82. 82. Alves-Gomes JA (1999) Systematic biology of gymnotiform and mormyriform electric fishes: phylogenetic relationships, molecular clocks and rates of evolution in the mitochondrial rRNA genes. J Exp Biol 202: 1167–1183.
  83. 83. Kumazawa Y, Nishida M (2000) Molecular phylogeny of osteoglossoids: a new model for Gondwanian origin and plate tectonic transportation of the Asian arowana. Mol Biol Evol 17: 1869–1878.
  84. 84. Peng ZG, He SP, Wang J, Wang W, Diogo R (2006) Mitochondrial molecular clocks and the origin of the major Otocephalan clades (Pisces: Teleostei): a new insight. Gene 370: 113–124.
  85. 85. Nakatani M, Miya M, Mabuchi K, Saitoh K, Nishida M (2011) Evolutionary history of Otophysi (Teleostei), a major clade of the modern freshwater fishes: Pangaean origin and Mesozoic radiation. BMC Evol Biol 11: 177.
  86. 86. Inoue JG, Kumazawa Y, Miya M, Nishida M (2009) The historical biogeography of the freshwater knifefishes using mitogenomic approaches: A Mesozoic origin of the Asian notopterids (Actinopterygii: Osteoglossomorpha). Mol Phylogenet Evol 51: 486–499.
  87. 87. Lavoué S, Miya M, Arnegard ME, McIntyre PB, Mamonekene V, et al. (2011) Remarkable morphological stasis in an extant vertebrate despite tens of millions of years of divergence. Proc R Soc Lond [Biol] 278: 1003–1008.
  88. 88. Wilson MVH, Murray AM (2008) Osteoglossomorpha: phylogeny, biogeography, and fossil record and the significance of key African and Chinese fossil taxa. In: Cavin L, Longbottom A, Richter M, editors. Fishes and the Break-up of Pangaea. London: Geological Society, Special Publications v295. pp. 185–219.
  89. 89. Taverne L (1998) Les Ostéoglossomorphes marins de l’Eocène du Monte Bolca (Italie): Monopteros Volta 1796, Thrissopterus Heckel, 1856 et Foreyichthys Taverne, 1979. Considérations sur la phylogénie des téléostéens ostéoglossomorphes. Studi e Ricerche sui Giacimenti Terziari di Bolca, Miscellanea Paleontologica Vol 7. Verona: Museo Civico di Storia Naturale. pp. 67–158.
  90. 90. Lavoué S, Bigorne R, Lecointre G, Agnèse JF (2000) Phylogenetic relationships of mormyrid electric fishes (Mormyridae, Teleostei) inferred from cytochrome b sequences. Mol Phylogenet Evol 14: 1–10.
  91. 91. Taverne L (1972) Ostéologie des genres Mormyrus Linné, Mormyrops Müller, Hyperopisus Gill, Myomyrus Boulenger, Stomatorhinus Boulenger et Gymnarchus Cuvier. Considérations générales sur la systématique des poissons de l’ordre des Mormyriformes. Ann Musee Roy Afr Centr Ser 8 Sci Zool 200: 1–194.
  92. 92. Wiley EO, Johnson GD (2010) A teleost classification based on monophyletic groups. In: Nelson JS, Schultze HP, Wilson MVH, editors. Origin and Phylogenetic Interrelationships of Teleosts. München: Verlag Dr. Friedrich Pfeil. pp. 123–182.
  93. 93. Cavin L, Longbottom A, Richter M, editors. (2008) Fishes and the break-up of Pangaea. London: The Geological Society, Special Publication. 396 p.
  94. 94. Patterson C (1993) Osteichthyes Teleostei. In: Benton MJ, editor. The fossil record. London: Chapman, et al. pp. 621–656.
  95. 95. Hurley IA, Mueller RL, Dunn KA, Schmidt EJ, Friedman M, et al. (2007) A new time-scale for ray-finned fish evolution. Proc R Soc Lond [Biol] 274: 489–498.
  96. 96. Yamanoue Y, Miya M, Inoue JG, Matsuura K, Nishida M (2006) The mitochondrial genome of spotted green pufferfish Tetraodon nigroviridis (Teleostei: Tetraodontiformes) and divergence time estimation among model organisms in fishes. Genes Genet Syst 81: 29–39.
  97. 97. Lynch M, Katju V (2004) The altered evolutionary trajectories of gene duplicates. TIG 20: 544–549.
  98. 98. Ohno S (1970) Evolution by gene duplication. New York: Springer-Verlag. 160 p.
  99. 99. Zhang JZ (2003) Evolution by gene duplication: an update. Trends Ecol Evol 18: 292–298.
  100. 100. Taylor JS, Raes J (2004) Duplication and divergence: The evolution of new genes and old ideas. Annu Rev Genet 38: 615–643.
  101. 101. Hoegg S, Brinkmann H, Taylor JS, Meyer A (2004) Phylogenetic timing of the fish-specific genome duplication correlates with the diversification of teleost fish. J Mol Evol 59: 190–203.
  102. 102. Sato Y, Nishida M (2010) Teleost fish with specific genome duplication as unique models of vertebrate evolution. Environ Biol Fishes 88: 169–188.
  103. 103. Novak AE, Jost MC, Lu Y, Taylor AD, Zakon HH, et al. (2006) Gene duplications and evolution of vertebrate voltage-gated sodium channels. J Mol Evol 63: 208–221.
  104. 104. Caputi AA, Carlson BA, Macadar O (2005) Electric organs and their control. In: Bullock TH, Hopkins CD, Popper AN, Fay RR, editors. Electroreception. New York: Springer Science+Business Media, Inc. pp. 410–451.
  105. 105. Briggs JC (2005) The biogeography of otophysan fishes (Ostariophysi: Otophysi): a new appraisal. J Biogeography 32: 287–294.
  106. 106. Li G-Q (1997) Notes on the historical biogeography of the Osteoglossomorpha (Teleostei). In: Jin Y-G, Dineley D, editors. Proc 30th Intern Geol Congr Vol 12, Palaeontol Hist Geol. Zeist: VSP BV. pp. 54–66.
  107. 107. Lundberg JG, Sullivan JP, Rodiles-Hernandez R, Hendrickson DA (2007) Discovery of African roots for the Mesoamerican Chiapas catfish, Lacantunia enigmatica, requires an ancient intercontinental passage. P Acad Nat Sci Phila 156: 39–53.
  108. 108. Sullivan JP, Lundberg JG, Hardman M (2006) A phylogenetic analysis of the major groups of catfishes (Teleostei : Siluriformes) using rag1 and rag2 nuclear gene sequences. Mol Phylogenet Evol 41: 636–662.
  109. 109. Barron EJ (1987) Cretaceous plate tectonic reconstructions. Palaeogeog Palaeoclim Palaeoecol 59: 3–29.
  110. 110. Scotese CR, Gahagan LM, Larson RL (1988) Plate tectonic reconstructions of the Cretaceous and Cenozoic ocean basins. Tectonophysics 155: 27–48.
  111. 111. Smith AG, Smith DG, Funnell M (1994) Atlas of mesozoic and cenozoic coastlines. Cambridge: Cambridge University Press. pp. 1–99.
  112. 112. Haq BU, Hardenbol J, Vail PR (1987) Chronology of fluctuating sea levels since the Triassic. Science 235: 1156–1167.
  113. 113. Miller KG, Kominz MA, Browning JV, Wright JD, Mountain GS, et al. (2005) The phanerozoic record of global sea-level change. Science 310: 1293–1298.
  114. 114. Jacobs LL, Mateus O, Polcyn MJ, Schulp AS, Scotese CR, et al. (2009) Cretaceous paleogeography, paleoclimatology, and amniote biogeography of the low and mid-latitude South Atlantic Ocean. B Soc Geol Fr 180: 333–341.
  115. 115. Maisey JG (2000) Continental break up and the distribution of fishes of Western Gondwana during the Early Cretaceous. Cretaceous Res 21: 281–314.
  116. 116. Fink SV, Fink WL (1981) Interrelationships of the ostariophysan fishes (Teleostei). J Linnean Soc [Zool] 72: 297–353.
  117. 117. DeVaney SC (2008) The interrelationships of fishes of the order Stomiiformes. Lawrence: University of Kansas. pp. 1–233.
  118. 118. Kawaguchi M, Hiroi J, Miya M, Nishida M, Iuchi I, et al. (2010) Intron-loss evolution of hatching enzyme genes in Teleostei. BMC Evol Biol 10: e260.
  119. 119. Lavoué S, Miya M, Inoue JG, Saitoh K, Ishiguro N, et al. (2005) Molecular systematics of the gonorynchiform fishes (Teleostei) based on whole mitogenome sequences: implications for higher-level relationships within the Otocephala. Mol Phylogenet Evol 37: 165–177.
  120. 120. Li CH, Lu GQ, Ortí G (2008) Optimal data partitioning and a test case for ray-finned fishes (Actinopterygii) based on ten nuclear loci. Syst Biol 57: 519–539.
  121. 121. Li J, Xia R, McDowall RM, Lopez JA, Lei GC, et al. (2010) Phylogenetic position of the enigmatic Lepidogalaxias salamandroides with comment on the orders of lower euteleostean fishes. Mol Phylogenet Evol 57: 932–936.
  122. 122. Santini F, Harmon LJ, Carnevale G, Alfaro ME (2009) Did genome duplication drive the origin of teleosts? A comparative study of diversification in ray-finned fishes. BMC Evol Biol 9: e194.
  123. 123. Dimmick WW, Larson A (1996) A molecular and morphological perspective on the phylogenetic relationships of the otophysan fishes. Mol Phylogenet Evol 6: 120–133.
  124. 124. Peng ZG, Wang J, He SP (2006) The complete mitochondrial genome of the helmet catfish Cranoglanis bouderius (Siluriformes : Cranoglanididae) and the phylogeny of otophysan fishes. Gene 376: 290–297.
  125. 125. Saitoh K, Miya M, Inoue JG, Ishiguro NB, Nishida M (2003) Mitochondrial genomics of ostariophysan fishes: perspectives on phylogeny and biogeography. J Mol Evol 56: 464–472.
  126. 126. Ortí G, Meyer A (1996) Molecular evolution of ependymin and the phylogenetic resolution of early divergences among euteleost fishes. Mol Biol Evol 13: 556–573.
  127. 127. Ortí G, Meyer A (1997) The radiation of characiform fishes and the limits of resolution of mitochondrial ribosomal DNA sequences. Syst Biol 46: 75–100.
  128. 128. Budd GE (2001) Why are arthropods segmented? Evol Dev 3: 332–342.
  129. 129. Löytynoja A, Milinkovitch MC (2003) A hidden Markov model for progressive multiple alignment. Bioinformatics 19: 1505–1513.
  130. 130. Stamatakis A (2006) RAxML-VI-HPC: maximum likelihood-based phylogenetic analyses with thousands of taxa and mixed models. Bioinformatics 22: 2688–2690.
  131. 131. Silvestro D, Michalak I (2011) raxmlGUI: a graphical front-end for RAxML. Org Divers Evol in press. DOI 10.1007/s13127–011–0056–0.
  132. 132. Tamura K, Peterson D, Peterson N, Stecher G, Nei M, et al. (2011) MEGA5: molecular evolutionary genetics analysis using maximum likelihood, evolutionary distance, and maximum parsimony methods. Mol Biol Evol 28: 2731–2739.
  133. 133. Yang Z (1994) Maximum likelihood phylogenetic estimation from DNA sequences with variable rates over sites: approximate methods. J Mol Evol 39: 306–314.
  134. 134. Yang Z (2006) Computational molecular evolution. Oxford: Oxford University Press. 376 p.
  135. 135. Drummond AJ, Rambaut A (2007) BEAST: Bayesian evolutionary analysis by sampling trees. BMC Evol Biol 7: e214.
  136. 136. Brandley MC, Wang YZ, Guo XG, de Oca ANM, Ferìa-Ortìz M, et al. (2011) Accommodating heterogenous rates of evolution in molecular divergence dating methods: An example using intercontinental dispersal of Plestiodon (Eumeces) lizards. Syst Biol 60: 3–15.
  137. 137. Lukoschek V, Keogh JS, Avise JC (2012) Evaluating fossil calibrations for dating phylogenies in light of rates of molecular evolution: a comparison of three approaches. Syst Biol 61: 22–43.
  138. 138. Phillips MJ (2009) Branch-length estimation bias misleads molecular dating for a vertebrate mitochondrial phylogeny. Gene 441: 132–140.
  139. 139. Benton MJ, Donoghue PCJ, Asher RJ (2009) Calibrating and constraining molecular clocks. In: Hedges SB, Kumar S, editors. The Timetree of Life. Oxford: Oxford University Press. pp. 35–86.
  140. 140. Nelson JS (2006) Fishes of the World. New York: John Wiley and Sons. 601 p.
  141. 141. Gardiner BG, Schaeffert B, Masserie JA (2005) A review of the lower actinopterygian phylogeny. Zool J Linnean Soc 144: 511–525.
  142. 142. Lavoué S, Miya M, Poulsen JY, Moller PR, Nishida M (2008) Monophyly, phylogenetic position and inter-familial relationships of the Alepocephaliformes (Teleostei) based on whole mitogenome sequences. Mol Phylogenet Evol 47: 1111–1121.
  143. 143. Heimberg AM, Cowper-Sallari R, Semon M, Donoghue PCJ, Peterson KJ (2010) microRNAs reveal the interrelationships of hagfish, lampreys, and gnathostomes and the nature of the ancestral vertebrate. Proc Natl Acad Sci, USA 107: 19379–19383.
  144. 144. Kikugawa K, Katoh K, Sakurai H, Ishida O, Iwabe N, et al. (2004) Basal jawed vertebrate phylogeny inferred from multiple nuclear DNA-coded genes. BMC Biol 2: e3.
  145. 145. Takezaki N, Figueroa F, Zaleska-Rutczynska Z, Takahata N, Klein J (2004) The phylogenetic relationship of tetrapod, coelacanth, and lungfish revealed by the sequences of forty-four nuclear genes. Mol Biol Evol 21: 1512–1524.
  146. 146. Andres KH, von Düring M, Petrasch E (1988) The fine structure of ampullary and tuberous electroreceptors in the South American blind catfish Pseudocetopsis spec. Anat Embryol (Berl) 177: 523–535.
  147. 147. Czech-Damal NU, Liebschner A, Miersch L, Klauer G, Hanke FD, et al. (2011) Electroreception in the Guiana dolphin (Sotalia guianensis). Proc R Soc Lond [Biol] published online ahead of print, 27 July 2011.
  148. 148. Forey PL (1997) A Cretaceous notopterid (Pisces: Osteoglossomorpha) from Morocco. South African J Sci 93: 564–569.