Salminus, Agassiz

Lima, Flávio C. T., 2022, Revision of the smaller-sized dorados (Salminus), with comments on the monophyly of the genus and its biogeography (Characiformes: Bryconidae), Zootaxa 5226 (1), pp. 1-66 : 5-11

Genus Salminus

Salminus Agassiz in Spix & Agassiz, 1829: 76. Type species: Hydrocyon brevidens Cuvier, 1819 . Type by monotypy. Gender: masculine.

Diagnosis. The genus Salminus can be recognized by the following unique conditions among characiforms: (1) fourth infraorbital very narrow and straight; (2) pedunculated teeth, with a cylindrical base and a roughly triangular crown; (3) sphenotic spine well developed, directed both laterally and forward. Additionally, the following characters are unique to Salminus within Bryconidae : (4) presence of an inner teeth row on dentary extending along the entire lingual crest of the replacement teeth trench; (5) frontoparietal fontanel anteriorly to the epiphyseal bar reduced to a small opening; (6) presence of a process contacting the anterior ceratohyal to the posterior ceratohyal. All these characters are discussed below.

Monophyly of Salminus . The monophyly of Salminus has never been questioned in the literature, as all species of the genus are very similar to each other morphologically, and recently, has been also supported by molecular data ( Abe et al., 2014; Machado et al., 2018). Mirande (2010: 478–479) listed 14 autapomorphies for S. brasiliensis in his phylogenetic morphological analysis of the family Characidae , from which only one is herein cited as a synapomorphy of Salminus (his character 181, articulation between anterior and posterior ceratohyals). Although many of the remaining autapomorphies may be present in the remaining species of Salminus , most of them are present in other Bryconidae and as such are not listed herein as unambiguous synapomorphies for the genus. One of the autapomorphies for S. brasiliensis listed by Mirande (2010), the presence of a foramen at the posterior region of the metapterygoid (his character 168, state 1; see Fig. 1A–B ), is actually variable within Salminus , with some species ( S. affinis , S. hilarii , and S. santosi ) presenting the state 2 (“in form of an incomplete arch, bordered posteriorly by hyomandibula”) ( Fig. 1C–D ). See further comments on the variation of this character in Salminus in the Discussion, below. A list of synapomorphies for the genus Salminus is presented below.

1. Shape of fourth infraorbital bone ( Vari & Harold, 2001: 27–28, character 31; Zanata & Vari, 2005: 11, character 13; Mirande, 2010: 408, character 67; Mattox & Toledo-Piza, 2012: 844: character 60; Ribeiro & Menezes, 2015: 44, character 10).

The fourth infraorbital bone is quite variable in shape among characiforms. Specifically for Bryconidae , its shape ranges from roughly square-shaped to rectangular (e.g., Weitzman, 1962: 64, fig. 8; Lima, 2017: 13, fig. 6). The fourth infraorbital bone in Salminus , although rectangular in shape, is much more elongated and narrower than in any other Bryconidae ( Fig. 2 ), and apparently, also unique among characiforms as a whole.

2. Teeth shape ( Buckup, 1998: 133, character 72; Mirande, 2010: 419, characters 118–119; Mattox & Toledo-Piza, 2012: 848–849, character 71).

Characiforms are renowned for possessing a wide range of variation in their teeth morphology, and the importance of this character in the systematics of the group has been compared to its importance in mammals ( Roberts, 1975). Characiform teeth can be either unicuspid or multicuspid ( Buckup, 1998). Multicuspidate teeth can be either pedunculate or presenting a broad base ( Mirande, 2010: 419). Pedunculated teeth present an elongated, either cylindrical or compressed basal portion (shaft) and an apical portion (crown). Among the Bryconidae , Brycon species possess multicuspidate teeth with a broad base (e.g., Howes, 1982: 23, fig. 15; Lima, 2017: 9, figs. 1–2), although some species possess relatively pointed teeth (e.g., B. insignis and B. alburnus ; Howes, 1982: 3, figs. 1–2). Compressed, elongated teeth that can be considered pedunculate in shape are found in the two monotypic genera, Chilobrycon and Henochilus (e.g., Géry & Rham, 1981: 7, fig. 2; Castro et al., 2004: 500, fig. 3, respectively). Teeth shape in Salminus has been considered as conical by some authors (e.g., Roberts, 1969: 436). However, as earlier remarked by Regan (1911: 18) the teeth of Salminus are not truly conical. They differ from the truly conical teeth found in several Characinae , and in Cynodontidae , Acestrorhynchidae , or Erythrinidae , among other characiforms, since they possess a noticeable crown, triangular in shape, and as such are best interpreted as pedunculated teeth ( Fig. 3A–B ). The only characiforms that present teeth that are morphologically reminiscent to the ones present in Salminus are the Ctenoluciidae (e.g., Ctenolucius hujeta ; Lawson & Manly, 1973: 386, fig. 4). The teeth in Ctenoluciidae however lack a distinctive crown and as such as better identified as conical, although considerably shorter and more robust than the ones present in the aforementioned characiforms presenting conical teeth. Teeth shape in Salminus can be consequently considered to be unique among characiforms. Cione & Azpelicueta (2013: 1056–1057) depicted and further discussed the morphology of the teeth in Salminus , but a detailed study of the teeth morphology in other characiforms is needed before some of the features listed by them as unique for the genus can be confirmed as such.

3. Sphenotic bone morphology ( Vari & Harold, 2001: 33–34, characters 43–44; Toledo-Piza, 2007: 706–707, character 19).

The sphenotic bone typically possess in Characiformes a lateral projection, called sphenotic spine. This projection is also formed by the frontal bone, although not distally. The sphenotic spine, although projecting laterally and being discernible in a lateral view of the neurocranium, is in the majority of the Characiformes direct- ed mainly downward and sometimes slightly backwards (e.g., Xenocharax spilurus: Vari, 1979: 284 , fig. 10; Pseudanos trimaculatus: Winterbottom, 1980: 102–104 , figs. 51–53; Ichthyoelephas humeralis: Roberts, 1973: 222–223 , figs. 2–4; Hemiodus semitaeniatus: Roberts, 1974: 437 , figs. 1–2; Hydrocynus forskahlii: Brewster, 1986: 167 , fig. 1; Hydrolycus armatus: Toledo-Piza, 2000: 14 , fig. 5; Brycon meeki ; Weitzman, 1962: 58–60, figs. 2–4; Tometes trilobatus: Jégu et al., 2002: 111 , fig. 12). The sphenotic spine is mainly directed laterally in some characiforms as Lebiasina bimaculata and Hepsetus odoe . In Salminus spp. , the sphenotic spine is well developed and directed mainly laterally and forward ( Fig. 4 ; see also Roberts, 1969: 462, fig. 19; Mattox et al., 2014: 111, fig. 5), a condition unique among characiforms.

4. Inner teeth row on dentary ( Malabarba, 1998: 75, character 41; Buckup, 1998: 130, character 36; Zanata & Vari, 2005: 46–47, character 88; Mirande, 2010: 425, character 143; Mattox & Toledo-Piza, 2012: 850, character 77). As remarked by Buckup (1998: 130), a second teeth row in the dentary, in Characiformes, when present, lies at the inner crest of the teeth replacement trench and is typically composed by small conical teeth. We here follow Lima (2017: 11–12) in considering the pair of inner symphyseal teeth, present in some characiforms, as a distinct character from the teeth row present at the inner crest of the teeth replacement trench. As herein defined, a second inner dentary inner row is present in Distichodontidae ( Vari, 1979: 276) , Microsmichodus sugillatus ( Hemiodontidae : Roberts, 1971), Crenuchidae (e.g., Buckup, 1993), Lebiasinidae ( Weitzman, 1964: 143) , some Erythrinidae ( Gayet et al., 2003) , Hepsetidae ( Roberts, 1969) , Chalceidae ( Zanata & Toledo-Piza, 2004: 105–106) , in some Acestrorhynchidae ( Mattox & Toledo-Piza, 2012: 850) , and in a few Characidae : some Characinae ( Mattox & Toledo-Piza, 2012: 850) , and Aphyocharacidium ( Géry, 1973) . In Bryconidae , a second dentary teeth row is seemingly present in all species, as it is present in Brycon ( Lima, 2017) , Henochilus wheatlandii ( Castro et al., 2004: 501) , and all Salminus species. However, the condition of the inner teeth row is distinct in Salminus from the one found in Brycon , as in the latter genus, the inner teeth row starts almost always considerably afterwards the symphysis, typically at the level of the fourth anteriormost teeth or after that point ( Lima, 2017), whereas in Salminus , it extends continuously from the symphysis to the end of the toothed portion of the dentary (e.g., Roberts, 1969: 469, fig. 31; Mirande, 2010: 426, fig. 71). Castro et al. (2004: 501) described the inner teeth row in Henochilus wheatlandii as being variable ontogenetically and absent from larger specimens, whereas no ontogenetic variation was detected in this character in any of the Brycon species studied by Lima (2017).

5. Fronto-parietal fontanels ( Vari, 1995: 15–16, character 26; Malabarba, 1998: 33, character 1; Buckup, 1998: 128, character 9; Zanata & Vari, 2005: 20–22, characters 36–37; Toledo-Piza, 2007: 705, character 15; Mirande, 2010: 395–396, character 22).

Most Characiformes present the frontal and parietal bones separated by a fontanel. The extension of the fontanel anteriorly to the epiphyseal bar is variable in some Characiformes, e.g., Distichodontidae ( Vari, 1979: 290) , Steindachnerina of the S. hypostoma species complex ( Curimatidae : Vari: 1991, fig. 10), and in the Alestidae ( Zanata & Vari, 2005: 20–22). The fontanel is absent in adult specimens of several characiform families, i.e., Ctenoluciidae , Erythrinidae , Hepsetidae , Lebiasinidae , and Parodontidae ( Vari, 1995: 15–16) . Within Bryconidae , the fontanel is normally developed (i.e., separating completely the frontals and parietals bones) in most Brycon , except for B. pesu , which lacks entirely a fontanel when adults (e.g., Weitzman, 1962; Howes, 1982; Lima, 2017). In all cleared and stained and skeletonized Salminus specimens examined in the present study, the fontanel is present and normally developed between the epiphyseal bar and the supraocippital but is reduced to a small opening anteriorly to the epiphyseal bar (e.g., Fig. 4 ; see also Roberts, 1969: 452, fig. 9). However, both Roberts (1969: 407) and Zanata & Vari (2005: 21) suggested that in large Salminus specimens the fontanel is occluded, and Mattox et al. (2014: 111, fig. 5) depicted a neurocranium of a specimen of S. brasiliensis in which the parietal fontanel was indeed partially occluded. The larger skeletons of the genus examined in the present study are of two specimens of S. brasiliensis ( MZUSP 89521, 437– 458 mm SL), both of which presented only a thin sheet of connective tissue over the parietal fontanel, which was not actually occluded. This is somewhat perplexing since the larger S. brasiliensis specimen depicted by Mattox et al. (2014) is smaller (332 mm SL) than the ones examined in the present study, but as their specimen was raised in captivity, it probably experienced a faster development than the wild specimens herein examined, a common condition for captive fishes (G.M. T. Mattox, pers. comm.). Although unfortunately no skeletons of large-sized Salminus specimens were available for examination, a cursorial examination of whole, wild-caught specimens of both S. brasiliensis and S. franciscanus ranging from 430–580 mm SL showed that the parietal fontanel is not occluded in any of them. However, the examination of a large S. franciscanus ( MZUSP 39000, 773 mm SL) was equivocal in that regard as the fontanel could not be discerned and might have indeed been occluded by hyperossification.

6. Articulation between ceratohyals ( Vari, 1995: 24–25, character 58; Mirande, 2010: 433, character 181; Mattox & Toledo-Piza, 2012: 858, character 97). The contact between the anterior and posterior ceratohyals is typically synchondral, but in a few characiforms, processes are present contacting both bones (e.g., Vari, 1995: 24–25; Mirande, 2010: 433; Mattox & ToledoPiza, 2012: 858). In Salminus , a process from the anterior ceratohyal contacts the posterior ceratohyal (see Roberts, 1969: 474, fig. 36).

The remaining Bryconidae , including all species examined in the present study (see Lima & Britski, 2007; Lima, 2017), lack any processes contacting both ceratohyals, and thus the condition found in Salminus can be considered a synapomorphy for the genus.

Phylogenetic relationships. The first explicit suggestion concerning the relationships of Salminus was by Regan (1911), who suggested that the genus was “closely related” to Hystricodon (= Exodon ) and Brycon , “from which they differ only in the dentition...” ( Regan, 1911: 18). Although a partial description of the osteology of S. maxillosus (= S. brasiliensis ) has been presented by Fuster de Plaza (1950) and Moraes & Schubart (1955), the first author to discuss the putative relationships of Salminus within Characiformes, based on an extensive study of its osteology, was Roberts (1969). Roberts (1969: 435) remarked that the overall resemblance of Salminus to Brycon was not necessarily an indicative of a close relationship between these genera and suggested (p. 442) that the genus might belong to a relictual lineage. Géry (1972: 50, 52, 55; 1977: 330–331, 334) considered Salminus as belonging to a tribe (Salminini, first proposed by Cockerell, 1915: 156) within the subfamily Bryconinae , which also included Brycon , Triportheus , Moojenichthys (= Lignobrycon ), Catabasis and Holobrycon (both currently considered as synonyms of Brycon ), without however presenting any diagnosis for the subfamily. The first published phylogenetic analysis of Characiformes that included the genus Salminus was by Ortí & Meyer (1997), which conducted a phylogenetic analysis using segments of the 12S and 16S subunit of the RNA mitochondrial genes. Ortí & Meyer (1997: 92) recovered a well-corroborated sister relationship between Salminus and Brycon . Malabarba & Weitzman (2003) considered Salminus , together with Brycon , Bryconops , Triportheus , Lignobrycon , and the representatives of Serrasalminae (currently Serrasalmidae ) and Gasteropelecidae , as a putatively basal Characidae , due to the presence of fin hooks on the pelvic and anal fins of mature males (a feature considered by the authors as a putative synapomorphy of the family), and the presence of a supraorbital bone (considered by them as a plesiomorphic feature within Characidae ). Lima et al. (2003) listed Salminus among the Characidae incertae sedis, and Lima (2003) restricted Bryconinae to the genera Brycon , Chilobrycon , and Henochilus . Calcagnotto et al. (2005) presented a broad phylogenetic hypothesis of Characiformes based on both nuclear and mitochondrial genes and obtained a similar result to Ortí & Meyer (1997) regarding the relationships of Salminus recovering it as a sister-group of Brycon . Mirande (2010) presented the first broad phylogenetic hypothesis of Characidae based exclusively on morphological data and recovered Salminus as the sister-group of Agoniatinae , Cynodontidae , and Acestrorhynchidae , all within a basal clade in Characidae that also included Bryconinae (which included, besides Brycon , also Triportheus ). The broad phylogenetic analysis of Characiformes based on molecular data by Oliveira et al. (2011), however, recovered Salminus as all previous molecular hypothesis did, as the sister-group of Brycon (recognized by these authors as the family Bryconidae ), and the clade formed by these two genera as sister to the family Gasteropelecidae , which in turn was recovered as the sister-group of the newly defined family Triportheidae . Although not including Salminus , Betancur-R. et al. (2018), based on a phylogenomic analysis, recovered Bryconidae as the sister-group of a clade including the families Iguanodectidae , Acestrorhynchidae , Gasteropelecidae , and Triportheidae . Abe et al. (2014) presented a phylogenetic molecular analysis of the family Bryconidae , composed by the genera Brycon , Chilobrycon , Henochilus , and Salminus . Finally, Mirande (2019) presented a total evidence phylogenetic analysis of Characidae , where he recovered a monophyletic Bryconidae with the same generic composition as Abe et al. (2014), but with a distinct intraordinal relationships than the one found in Oliveira et al. (2011) or Betancur-R. et al. (2018), as Bryconidae was recovered as the sister group of a clade including Gasteropelecidae , Triportheidae , and Characidae .

The relationships of Salminus within Bryconidae were addressed by Abe et al. (2014), who surprisingly recovered the species of the genus as forming a clade within the genus Brycon , being the sister-group of a clade containing almost exclusively cis-andean species of Brycon , plus the monotypic Henochilus , and the clade formed by this latter clade and Salminus having as its sister-group a clade formed exclusively by trans-andean species of Brycon , plus the monotypic Chilobrycon . Mirande (2019), based on a less extensive sampling of the family, obtained a similar result. As for the intrarelationships of Salminus , there are two published hypotheses, both based on molecular data. The first, by Abe et al. (2014), recovered S. hilarii as the basalmost taxon, sister to a clade formed by two clades, one composed by S. affinis and the Salminus species from the rio Tocantins basin (herein described as S. santosi ), and the other composed by S. brasiliensis and S. franciscanus . The second, by Machado et al. (2017, 2018), recovered S. affinis as the basalmost taxon of the genus, with two successive main clades, the first composed by “ Salminus sp. Amazon (Upper Amazon)” plus “ Salminus sp. Amazon (Madeira)” (= S. iquitensis ) being sister to “ Salminus sp. Araguaia” (= S. santosi ), and the second with S. franciscanus as the sister group of a clade formed by S. brasiliensis and S. hilarii . Machado et al. (2018: 1320) suggested that the discrepancy between their results when compared with the one presented by Abe et al. (2014) was caused by the mislabeling of the tissue sample identified by the later authors as S. hilarii , which would correspond rather to a species of Brycon , according to a reanalysis of the sequences made by them. This explanation is however a bit perplexing given the fact that Abe et al. (2014) included a very broad sampling of Brycon species and still the sequences identified by them as belonging to S. hilarii grouped with Salminus and not with any Brycon species. Both hypotheses are discussed regarding their systematic and biogeographical implications in the “Discussion”, below.

Figures

FIGURE 1. Metapterygoid bone in different Salminus species, showing the presence of a foramen (A–B) or a concavity (C–D) on the posterior margin of the bone. A. Salminus franciscanus, MZUSP 89512, 225.0 mm SL; B. Salminus iquitensis, MZUSP 25949, 119.1 mm SL; C. Salminus. hilarii, MZUSP 40113, 121.4 mm SL; D. Salminus santosi, MZUSP 53620, 154.8 mm SL. Anterior to right. Scale bar: 1 mm (A–C); 2 mm (D).

FIGURE 2. Infraorbital series of Salminus brasiliensis, MZUSP 89521, 437 mm SL, infraorbital series, showing the elongated fourth infraorbital. Abbreviations: ANT, antorbital; i01–06, infraorbitals one to six; SUP, supraorbital. Scale bar: 20 mm.

FIGURE 3. Middle portion of the dentary of Salminus santosi (MZUSP 53620, 154.8 mm SL), showing teeth shape in outer (A) and inner (B) views. Anterior to left (A) or to right (B).

FIGURE 4. Neurocranium of Salminus brasiliensis, MZUSP 89521, 437 mm SL, in dorsal view, showing the forward-directed sphenotic (SPH) and the fontanel (FNT) restricted to posterior to the epiphyseal bar (EPB). Other abbreviations: EPIOC: epioccipital; FR: frontal; MES: mesethmoid; NAS: nasal; PMX: premaxillary; PAR: parietal; PTE: pterotic; SUPR: supraoccipital. Scale bar: 50 mm.