|A phylogenetic study of the African catfish family Mochokidae (Osteichthyes, Ostariophysi, Siluriformes), with a key to its genera
Thomas R. Vigliotta
Cornell University Museum of Vertebrates, 159 Sapsucker Woods Road, Ithaca, NY 14850-1923, USA. email: firstname.lastname@example.org
ABSTRACT– A hypothesis of phylogenetic relationships, based on maximum parsimony analysis of 92 morphological characters, is presented for the Mochokidae for the first time. Forty-two ingroup taxa are considered in a heuristic search, and a subset of 17 taxa in a smaller exhaustive search. The results of the two analyses are largely in agreement and reveal the following: the Mochokidae are a monophyletic group, sister to a group composed of the South American Doradidae + Auchenipteridae; new synapomorphies for the family include features of the pectoral girdle, caudal fin and mandibular sensory canal; monotypic Mochokiella, Acanthocleithron and Atopodontis are valid as separate genera; Mochokus and Microsynodontis are monophyletic; Synodontis must include Hemisynodontis membranaceous and Brachysynodontis batensoda to be monophyletic; Chiloglanis is rendered paraphyletic by nested placement of Atopochilus, Euchilichthys and Atopodontis; and Euchilichthys is rendered paraphyletic by nested placement of Atopochilus savorgnani. Monophyly of Atopochilus was not tested because only one species was available. The recovered topology presents the following well-supported groups: Atopochilus and Euchilichthys form a monophyletic group, despite the paraphyly of Euchilichthys; Atopodontis, Atopochilus and Euchilichthys form a monophyletic group; Chiloglanis, Atopodontis, Atopochilus and Euchilichthys form a monophyletic group, despite the paraphyly of Chiloglanis; that group delimits the redefined subfamily Chiloglanidinae; the Chiloglanidinae form a trichotomy with Synodontis and Microsynodontis. Remaining intrafamilial groups are based on the placement of Mochokus, Mochokiella and Acanthocleithron towards the base of the tree; support for these groups is poor. Synapomorphies are listed for each major clade recovered. Well-supported groups within the Mochokidae are largely defined by a combination of characters pertaining to the teeth, oral jaws, and suspensorium. Changes in the jaw and suspensorium are a key theme in mochokid evolution. The results of this analysis indicate the need for the revision of several mochokid groups. Higher-level systematics of the Mochokidae can be greatly improved with continued emphasis on broader, lower-level taxonomic work.
ADDITIONAL KEYWORDS: Chiloglanis – morphology – osteology – phylogeny – squeakers – Synodontis – systematics – upside-down catfish.
The Mochokidae are a family of African catfishes distributed throughout the freshwaters of Africa. Commonly known as squeakers or upside-down catfishes, they are most well known for their appearance in the pet trade; in truth, most species found in the pet trade belong to a single large genus, Synodontis. Not surprisingly, the common names for the family are largely attributed to species of Synodontis. Species in the genus are capable of making a squeaking noise by stridulation of the pectoral spine against the pectoral girdle (Jubb, 1967); stridulation is also apparent in Mochokiella paynei and species in the genus Atopochilus (personal observation). Even more peculiar is the habit of some species of Brachysynodontis and Synodontis that are known to swim upside-down. This habit seems to be correlated with feeding while upside-down at the water’s surface (Bishai & Abu Gideiri, 1965), but upside-down catfishes will rest and swim in the inverted position on a regular basis. Chapman et al. (1994) showed that an upside-down posture near the surface also facilitates respiration in poorly oxygenated water. While these common names may refer to members of the genus Synodontis, the family is actually much more interesting when taken as a whole.
Species in the family are quite variable in terms of morphology. Species of Synodontis, Hemisynodontis, Brachysynodontis and Microsynodontis are known for their ventrally directed mouths, S-shaped teeth, branched mandibular barbels, well developed nuchal shields and bold pigmentation patterns. Species of Chiloglanis, Atopochilus, Euchilichthys and Atopodontis are rather drab in comparison and exhibit a distinctive, ventrally-directed oral disc. Lesser known mochokids, like Mochokus, Mochokiella and Acanthocleithron exhibit unique morphologies of their own. Size is also quite variable within the family. Altogether, mochokids range in size from 40mm SL in some species of Chiloglanis and Microsynodontis up to 800mm SL in some species of Synodontis. In addition to morphological variability, mochokids are known to exhibit obvious sexual dimorphism. For example, several species in the genus Chiloglanis show dimorphism of the anal and caudal fins (Roberts, 1989; Seegers, 1996; Friel & Vigliotta, 2006); some Chiloglanis also display sexual dimorphism of the cleithral process, wherein males possess a greatly enlarged process shielding the flank. In the genus Atopochilus, sexual dimorphism of the anal fin is sometimes evident (personal observation). Some mochokids also exhibit spiny ornamentation of the skull roof bones, opercular series and pectoral girdle (Friel & Vigliotta, 2006). In the case of Synodontis acanthoperca, a spine found at the rear of the opercle is, itself, sexually dimorphic. The spines of males are much larger than those of females. This is also true for Mochokiella paynei, which was previously unknown to possess opercular spines or exhibit sexual dimorphism.
The greatest diversity of mochokids occurs in the Congo River and its numerous tributaries, but they are also found in the rivers and lakes of western Africa, southern Africa, and eastern Africa, including the Nile River, Lake Victoria, Lake Tanganyika and Lake Nyasa. Like a handful of other catfishes, some mochokids are known to swim in mid-water; other members of the family are primarily benthic. Likewise, some mochokids shoal while others are rather solitary. As a rule they are most active during the night, but they can be found hiding amongst plants, logs and other submerged structure during the day. Fossil mochokids, of the genus Synodontis, have been found in deposits from eastern and northern Africa dating to the early Miocene (at least 20 mya) (Stewart, 2001). Fragments of pectoral spines from Synodontis dating from the early Oligocene have been found in Oman, an area where mochokids do not exist today (Otero and Gayet, 2001). Fossil mochokids outside of the genus Synodontis are presently unknown.
To date, nearly 250 species of mochokid catfishes have been described, but a large number of those names are now considered junior synonyms. The number of valid, described species is approaching 200 and several undescribed species are known to exist. This makes the Mochokidae one of the largest families of catfishes and the largest family of African catfishes by far. Ferraris (2007) provides the most recent list of mochokid species and synonymies (188 listed as valid). Synodontis acanthoperca Friel and Vigliotta (2006), Chiloglanis productus Ng and Bailey (2006) and three species of Synodontis from Lake Tanganyika (Wright and Page, 2006) were described too recently to be included in that work; Wright and Page also resurrect two species of Synodontis from synonymy. Two new mochokid species are also described in separate papers within this special volume (Friel and Vigliotta, 2008; Friel and Sullivan, 2008). In all, nearly one third of African catfishes belong to the family Mochokidae.
As with most African freshwater fishes, phylogenetic treatments of African catfishes are considerably rare. Taxonomic revisions of the African Schilbeidae (De Vos, 1995) and Malapteruridae (Norris, 2002) have been undertaken, but the only African group treated in a phylogenetic framework is the Amphiliidae (He et al., 1999; Diogo, 2003). For the several other families of catfishes inhabiting the African continent there has been very little phylogenetic work done. Beyond some recent work on the Synodontis from Lake Tanganyika (Koblmuller et al., 2006; Day and Wilkinson, 2006), phylogenetic treatment of the Mochokidae is practically non-existent.
The objectives of the this research were threefold and prioritized as such: (1) to assess the monophyly of known mochokid genera and to determine the phylogenetic relationships between these genera, (2) to assess the monophyly of the Mochokidae and (3) to support/establish a clade of catfishes as the sister group to the Mochokidae. I have approached these objectives by performing a broad survey of the gross morphology of several mochokid species in addition to several species serving as outgroups. A data matrix based on this survey was analyzed using parsimony methods in order to reconstruct a phylogeny for these taxa.
TAXONOMIC AND SYSTEMATIC HISTORY OF THE MOCHOKIDAE
The family Mochokidae was established long after the description of many of its component taxa; the history of the family is, as a result, fairly complicated. What follows is a mostly chronological account of the taxonomic history of the family and its genera.
The first mochokid species (Synodontis clarias today) was described by Linnaeus (1758) and placed in the genus Silurus. Cuvier (1816) described the first mochokid genus, Synodontis. He recognized it as distinct from other catfishes by placing it in his group “Des Schals” and characterized the group by the form of the teeth, among other features. The presence of S-shaped teeth remains an important characteristic of many mochokids today. Presently, there are approximately 120 species in the genus Synodontis. To date, no substructure has been proposed for the genus and its monophyly has never been demonstrated. Poll (1971) provides the most recent revision of the genus and a key to the species. The geographic distribution of Synodontis is similar to that of the family as a whole; they are found in many rivers and lakes throughout Africa.
Joannis (1835) described a second genus, Mochokus, but did not ally it with Synodontis. Joannis was justified in missing or ignoring a relationship to Synodontis because, unlike most mochokids, Mochokus does not have S-shaped teeth or a highly modified jaw. Today there are just two valid species in the genus Mochokus. Their geographic distribution is limited to northern and western Africa, including the Nile River.
Bleeker (1862-1863) reviewed several species of Synodontis that had been described in the intervening years and created four new genera to accommodate certain species. Leiosynodontis and Pseudosynodontis (both monotypic) were never widely used and were effectively synonymized by Günther (1864). Brachysynodontis and Hemisynodontis remain today as monotypic genera. The validity of these two genera is questionable at best. Some authors have already suggested that the two species, H. membranaceous and B. batensoda, belong in the genus Synodontis (Willoughby, 1994). The status of Hemisynodontis and Brachysynodontis will be effectively settled in this work. Both species are found in northern and western Africa.
The genus Rhinoglanis, now considered a junior synonym of Mochokus, was described by Günther (1864). Günther recognized that Rhinoglanis and Mochokus were allied to, but separate from Synodontis by placing each of them in his family Siluridae Stenobranchiae, but in different “groups.” He placed Rhinoglanis and Mochokus in his “group Rhinoglanina,” with a member of the Asian family Sisoridae. He placed Synodontis in the “group Doradina,” which also contained members of the South American families Cetopsidae, Auchenipteridae and Doradidae. Günther’s placement of Synodontis is noteworthy because the Auchenipteridae and Doradidae have been recovered as close relatives of the Mochokidae in more recent works (Chardon, 1968; Mo, 1991; de Pinna, 1993; Lundberg, 1993; Diogo, 2005). Günther’s “group Doradina” may constitute the first hypothesis of a close relationship between mochokids and these Neotropical families. One significant aspect of this relationship lies in the implied biogeographical scenario. Among catfishes, this is the only trans-Atlantic clade with solid support, morphological or otherwise. Lundberg (1993) suggested that this relationship is the result of an African-South American vicariance event.
Peters (1868) described the sucker-mouthed genus Chiloglanis and allied it to Günther’s “group Doradina.” Today Chiloglanis is the second largest mochokid genus, with approximately 45 valid species. Several undescribed species are known to exist. Again, no substructure within the genus has been proposed. The geographic distribution of Chiloglanis is similar to that of the family as a whole, although they are strictly riverine species and absent from lakes.
When Sauvage (1879) described a second sucker-mouthed genus, Atopochilus, he incorrectly allied it to Günther’s “group Ariina” in the family Siluridae Proteropterae. Today there are seven valid species of Atopochilus. The distribution of Atopochilus species is peculiar. The type species is described from the Lower Guinea ichthyofaunal province. Five of the remaining species are restricted the Congo basin and the final species is described from eastern Africa, in the Wami River of Tanzania.
Boulenger (1900) made an effort to bring all of the components, described thus far, together in a paper concerning the so-called subfamily Doradinae. He allied Chiloglanis, Atopochilus, and Mochokus with Synodontis, and synonymized Rhinoglanis with Mochokus. He also described and allied Euchilichthys, a new sucker-mouthed genus, to the group. Today there are five valid species in the genus Euchilichthys, though the distinction between Euchilichthys and Atopochilus has always been unclear. Between the two, a small number of undescribed species are known to exist. Euchilichthys species are limited to the Congo basin.
In a few short years Boulenger (1903) described the first species in the genus Microsynodontis, immediately allying it with Synodontis. Today there are 11 valid species in the genus Microsynodontis; eight of those species were described recently (Ng, 2004). At least one undescribed species is known and the genus is probably significantly larger than expected. Microsynodontis species can be found in the Congo basin, the Lower Guinea ichthyofaunal province and western Africa.
In his classification of teleostean fishes, Regan (1911) placed the mochokid genera into their own family for the first time, the Synodontidae. He also noted the fundamental differences between the sucker-mouthed varieties (Chiloglanis, Atopochilus, and Euchilichthys) and the other genera (Synodontis, Mochokus, Microsynodontis). Regan made no mention of Hemisynodontis or Brachysynodontis. Later, Jordan (1923) proposed the family name Mochokidae because the name Synodontidae was preoccupied by a family of marine lizardfishes, though the issue was confused for decades to come. Russell (1987) provides an account of the confusion surrounding this issue.
Nichols and Griscom (1917) described the first and only species in the genus Acanthocleithron. They recognized it as being within Boulenger’s and Regan’s mochokid groups, and allied it with Mochokus. A total of ten specimens are known for this very distinct species, yet it has a surprisingly large geographical range throughout the Congo basin.
In a detailed review of anatomy, myology and osteology of the genus Synodontis, Taverne and Aloulou-Triki (1974) created a subfamily, the Simuldentinae, containing Synodontis, Brachysynodontis, and Hemisynodontis. While there is no obvious reason to question the monophyly of this group, the act is curious because they did not treat any other genera within the family. According to Ferraris (2007) the Simuldentinae is not a valid subfamily or ‘family-group name’ because it is not based on an available genus name.
Another family group is recognized from a name proposed in the aquarium literature (Riehl and Baensch, 1990). The subfamily name Chiloglanidinae is applied to two species of Chiloglanis in that work, but the name is most appropriate for a group composed of all sucker-mouthed species. This subfamily name and the term ‘chiloglanidin’ will be used throughout the text to refer to a group composed of Chiloglanis, Atopochilus, Euchilichthys, and Atopodontis. The monophyly of this group will be tested by the analyses in this work.
Howes (1980) described the first and only species in the genus Mochokiella from Sierra Leone. Howes was unwilling to make any specific statement about the placement of this new genus, but allied it most closely with Microsynodontis. Mochokiella remains a monotypic genus. As far as known, Mochokiella is limited to small coastal drainages in Sierra Leone.
The newest mochokid genus, Atopodontis (Friel and Vigliotta, 2008), is described in this volume. Atopochilus adriaensi is a sucker-mouthed species that shares a number of features with species of Chiloglanis, but also Atopochilus and Euchilichthys. The placement and validity of the genus is tested in this work. Thus far, Atopodontis is known from the Ogôoué and Nyanga Rivers in Gabon.
The application of molecular phylogenetic techniques has resulted in new hypotheses as to the placement of the Mochokidae within the Siluriformes, but relationships within the family are still unknown. Recent work suggests that the Mochokidae are related to several African catfish families and that they are most closely related to the Amphiliidae (loach catfishes) and Malapteruridae (electric catfishes) (Sullivan et al., 2006). These relationships are well supported by the molecular data, but they greatly conflict with the phylogenetic scheme supported by morphology.
To summarize, the intrafamilial relationships for the Mochokidae have never been sufficiently examined. Original authors have allied genera based on gross morphological similarities, but never in a discrete phylogenetic context. Molecular phylogenies have considered very few mochokid taxa at this point and do not provide a comprehensive picture of the family’s diversity or evolutionary history. The relationship of the Mochokidae to other catfishes is also unclear. Morphological and molecular phylogenies have not recovered the same group(s) as sister to the Mochokidae and, in fact, the hypotheses are drastically different.
MATERIALS AND METHODS
Materials and Preparations
Materials examined in this study were prepared as cleared and stained specimens, dry skeletons, or fluid specimens. Appendix 1 lists the material examined and denotes the family, species, catalog number, number of specimens by each preparation type, country and specific locality. For preparation type, C&S = cleared and stained, DS = dry skeleton and ALC = alcoholic, fluid specimen. The materials are listed alphabetically by family, genus and species. Institutional abbreviations follow Leviton et al. (1985) except for the South African Institute of Aquatic Biodiversity (SAIAB).
Specimens were cleared and stained using the techniques outlined by Potthoff (1984). Generally, two specimens from the same lot were cleared and stained in order to account for at least some variation within a species, to examine sexually dimorphic characters and to rule out anomalous features. Dry skeletons were prepared with a dermestid beetle colony after cleaning by hand. External morphological features were examined on fluid specimens, but also on cleared and stained specimens when appropriate. Fluid specimens and cleared and stained specimens were usually available from the same lot.
Because material for the genus Acanthocleithron is exceedingly rare, other techniques were required to glean the necessary osteological information. A paratype specimen of Acanthocleithron chapini (AMNH 6575) was sent to the University of Texas at Austin to undergo High-Resolution X-ray Computed Tomography. High-resolution X-ray CT is a completely nondestructive technique for visualizing features in the interior of opaque solid objects, and for obtaining digital information on their 3-D geometries and properties. It is useful for a wide range of materials, including bone and soft tissue. Individual slices and video reconstructions of the three-dimensional specimen allowed me to visualize many of the internal features not otherwise accessible in this rare species.
Delimitation of Taxa
The ingroup considered in this work is the family Mochokidae, which includes nearly 200 valid species. Fifty-five ingroup species were initially examined (at least three undescribed), with at least one species from all of the valid genera within the family. The following taxa were considered valid genera within the family; numbers in parentheses represent the number of species examined: Acanthocleithron (1 valid species), Atopochilus (1 valid species), Atopodontis (1 valid species), Brachysynodontis (1 valid species), Chiloglanis (8 valid species, 3 undescribed, 1 unidentified), Euchilichthys (2 valid species, 2 unidentified), Hemisynodontis (1 valid species), Microsynodontis (2 valid species, 1 unidentified), Mochokiella (1 valid species), Mochokus (2 valid species), Synodontis (28 valid species). For Synodontis and Chiloglanis, the large number of described species required careful selection. In both instances, species were chosen to maximize the range of morphological variation within the genus and to cover as much of the known geographic range as possible. Type species were examined for all of the ingroup genera except Chiloglanis, Euchilichthys and Microsynodontis; suitable material for those type species was not available. All type species examined were included in the most comprehensive analysis. A small number of undescribed and otherwise unidentifiable species were included in the analyses because of their unique morphologies. Within this work, they are referred to by an informal name (often a locality) in quotation marks.
Outgroup taxa include members of several families of catfishes, including some putatively basal groups and some alleged to be closely related to the Mochokidae. Thirty-seven species from the following families or genera were examined for the reasons provided:
Diplomystidae – The Diplomystidae is the sister group to all other extant catfishes according to many authors (Alexander, 1965; Fink and Fink, 1981; Arratia, 1987; Mo, 1991; de Pinna, 1993). The family was recently established as sister to all siluroid catfishes, with loricarioids as sister to that group (Sullivan et al., 2006). In either case, because loricarioids are not considered here, the Diplomystidae are sufficiently basal with respect to the Mochokidae and all of the other outgroups. Data for the Diplomystidae were taken from previous studies dealing with higher-level siluriform relationships (Mo, 1991; de Pinna, 1993) and the works of Arratia (1987; 1992) and Arratia and Huaquin (1995).