ANCISTRINI, Kner, 1853

Armbruster, Jonathan W., 2004, Phylogenetic relationships of the suckermouth armoured catfishes (Loricariidae) with emphasis on the Hypostominae and the Ancistrinae, Zoological Journal of the Linnean Society 141 (1), pp. 1-80 : 54-57

ANCISTRINI

Isbrücker (1980) and Schaefer (1986, 1987) diagnosed the Ancistrini (then the Ancistrinae ) on the basis of the presence of evertible cheek odontodes and/or characteristics associated with them; however, as mentioned above, evertible cheek odontodes are also found in the Pterygoplichthini and support the sister-group relationship of the Ancistrini and Pterygoplichthini . Schaefer (1986, 1987) further diagnoses the Ancistrini by the presence of a derived opercle (75: 1/2); however, Hemiancistrus sp. Brazil that is the sister to all other members of the Ancistrini has an unmodified opercle (75: 0). With evertible cheek plates and modified opercles no longer able to diagnose the Ancistrini , the Ancistrini is left with no significant synapomorphies (Appendix 3) and it is appropriate to place Ancistrinae into the synonymy of Hypostominae .

Within the Ancistrini , several taxonomic problems are inherent. No characteristics are found to suggest that Cordylancistrus is monophyletic ( Fig. 38 ). The basic difference between species of Chaetostoma , Cordylancistrus , Dolichancistrus , and Leptoancistrus is the lack of plates along the snout of Chaetostoma ; otherwise, the species are similar. Given the phylogeny, there are several possibilities of how to make the taxonomy reflect phylogeny, including describing a new genus for Cordylancistrus platyrhynchus (Fowler) , placing it in Chaetostoma , or placing Cordylancistrus , Dolichancistrus , and Leptoancistrus into the synonymy of Chaetostoma . I act conservatively and place C. platyrhynchus in Chaetostoma with the genus diagnosed by the following characteristics: loss of suture between the pterotic-supracleithrum and hyomandibula (34: 0), loss of the hyomandibula angled mesially so that the opercle is held almost perpendicular to the main body axis (42: 0), the anterior process of the pterotic-supracleithrum is slightly deflected mesially (111: 0), reversal to narrow ventral process of sphenotic (116: 0), and tip of transverse process of the complex centrum of the Weberian apparatus not contacting the pterotic-supracleithrum (135: 1).

Although I have not examined Lipopterichthys osteologically, the genus is virtually indistinguishable from Chaetostoma except for the lack of adipose and anal fins (one species of Chaetostoma , C. venezuelae Schultz , shares the loss of the adipose fin). There are several specimens of Chaetostoma at the Auburn University Museum from near the type locality of Lipopterichthys carrioni (AUM 28213, 28215, 28222, and 28227). These are variable in the presence of the adipose fin and some have very reduced anal fins. It is probable that the type of Lipopterichthys is simply a morphotype of this variable species of Chaetostoma ; therefore, Lipopterichthys is recognized as a synonym of Chaetostoma .

Of the genera of the Hypostominae described in Isbrücker et al. (2001), the only one that can be adequately supported as a genus is Pseudolithoxus Isbrücker and Werner. Pseudolithoxus was described as the Lasiancistrus anthrax group of Armbruster & Provenzano (2000), although they provided little diagnostic information. Isbrücker et al. (2001) only repeat Armbruster and Provenzano’s diagnosis. Pseudolithoxus anthrax lacks most of the synapomorphies for Lasiancistrus and shares with other Lasiancistrus only the derived presence of a bifurcated anterior process of the pterotic-supracleithrum (113: 1). In addition, P. anthrax lacks synapomorphies of Lasiancistrus + Ancistrus , including the presence of tentacules on the snout larger than the supporting odontodes and tentacules on the pectoral-fin spines ( Sabaj et al., 1999). Pseudolithoxus shares with Pseudancistrus the trait of both males and females developing hypertrophied odontodes along the snout anterior to the cheek. Both males and females also develop extremely hypertrophied but flexible odontodes on the pectoral-fin spine of a type seen elsewhere only in Lithoxus . Given the phylogeny, hypertrophied odontodes along the snout in males and females and extremely hypertrophied odontodes on the pectoral-fin spines could be used as synapomorphies for Pseudolithoxus ; however, there are no other characteristics that serve to diagnose the genus.

Lasiancistrus is almost certainly a polyphyletic genus. Heitmans, Nijssen & Isbrücker (1983) describe several species that appear to be unrelated to the type species of Lasiancistrus [ L. heteracanthus (Günther) ]. In addition, an examination of type specimens of the species of Lasiancistrus sensu Isbrücker (1980) reveals several that more properly should be placed in other genera (including Chaetostoma , Hemiancistrus , Peckoltia , and Pseudancistrus ). Lasiancistrus should be restricted to those species with three rows of plates on the caudal peduncle and the presence of whiskerlike odontodes on the evertible cheek plates. Lasiancistrus will be detailed in a future publication on the Ancistrini , and the species that do not belong in Lasiancistrus will be discussed then.

Hemiancistrus sensu Isbrücker (1980) is polyphyletic with several species representing the sister to Pterygoplichthys as mentioned above, and the remain- der in several clades of the Ancistrini ( Fig. 38 ). The type species of Hemiancistrus , H. medians , is clearly a member of the Ancistrini , although no specimens are available for osteological examination. Based on superficial examination it appears that H. medians is closely related to Peckoltia . Three taxa clearly unrelated to Hemiancistrus s.s. are Hemiancistrus sp. Brazil, H. landoni , and H. megacephalus . Hemiancistrus sp. Brazil lacks the modified opercle diagnostic for the remainder of the Ancistrini . It is likely that a new genus needs to be described for Hemiancistrus sp. Brazil and possibly other south-eastern Brazilian Hemiancistrus (see Cardoso & Malabarba, 1999). The distribution of H. landoni is restricted to the Gulf of Guayaquil drainage of western Ecuador. No other Hemiancistrus ( H. hammarlundi is also described from west of the Andes, but is a synonym of H. landoni , pers. observ.) or the phenetically similar Peckoltia occur to the west of the Andes. H. megacephalus was well-supported as sister to Pseudancistrus and is transferred to Pseudancistrus .

Peckoltia is very problematic and is polyphyletic in this analysis ( Fig. 38 ). In addition, it is likely that the type species of Hemiancistrus ( H. medians ) is related to Peckoltia . Because the relationships of Peckoltia are not resolved and because there is the potential that some species of Hemiancistrus may be congeneric with species of Peckoltia , no changes to the taxonomy are made. Before the taxonomic problems inherent in Peckoltia and Hemiancistrus can be solved, Peckoltia and Hemiancistrus must first be revised. Isbrücker et al. (2001) describe Sophiancistrus for P. ucayalensis ; however, this study is not conclusive on the relationships of P. ucayalensis . Until such time as a future study can conclusively determine whether Sophiancistrus should be recognized, I recognize Sophiancistrus as a synonym of Peckoltia .

Armbruster (2002) recognized a new species of Hypancistrus and suggested that Hypancistrus could be diagnosed by bent adductor palatini crest and loss of the anterior contact of the metapterygoid and lateral ethmoid. In this study, Hypancistrus was not recovered as monophyletic; however, there is little support for relationships of species closely allied with Peckoltia (such as those of Hypancistrus and Parancistrus ) in the analysis and there is no justification for splitting Hypancistrus at this time.

Schaefer & Stewart (1993) provide compelling evidence that Panaque is monophyletic, although they did not examine Scobinancistrus which was found to be the sister to P. nigrolineatus in this analysis ( Fig. 38 ). Scobinancistrus shares with Panaque the presence of tall ridges on the hyomandibula and preopercle, a characteristic they listed as a synapomorphy for Panaque . This ridge is much taller in Panaque and Scobinancistrus than in most other loricariids (44), but I considered it too subjective to include an additional state of the levator arcus palatini crest in this analysis. Schaefer and Stewart also list the presence of an elongate, narrow metapterygoid channel as a synapomorphy for Panaque . Scobinancistrus lacks the lateral wall of the pterygoid channel and, hence, the state cannot be homologized with that seen in Panaque . The third synapomorphy for Panaque (hypertrophied muscles between the jaw rami) was not examined in Scobinancistrus due to lack of materials. Scobinancistrus differs from Panaque mainly in the lack of true spoon-shaped teeth; however, the teeth in Scobinancistrus appear as if they might be elongated spoon-shaped teeth ( Fig. 35D vs. C).

Isbrücker et al. (2001) describe Panaquolus as a new genus for the small members of Panaque (such as P. maccus and P. albomaculatus in this study); however, Chockley & Armbruster (2002) placed Panaquolus in the synonymy of Panaque , stating that there was no reason to accept Panaquolus as valid. In order for the taxonomy to reflect phylogeny, either Panaquolus must be recognized or Scobinancistrus placed into the synonymy of Panaque . In order to make the taxonomy reflect phylogeny more effectively it is better to recognize larger genera and to break them down into subgenera; thus, I retain Panaquolus as a synonym of Panaque , place Scobinancistrus into the synonymy of Panaque , and recognize three subgenera ( Panaque, Panaquolus , and Scobinancistrus ) in Panaque .

Oligancistrus Rapp Py-Daniel and Spectracanthicus are supported as sister taxa ( Fig. 38 ) by the mesial wall of the metapterygoid being much taller than the lateral wall (55: 1), a spoon-shaped anterior process of the metapterygoid (58: 1; Fig. 15D ), a deep pouch on the lateral ethmoid (98: 1; Fig. 20A ), and expansion of the dorsal-fin membrane such that it contacts the preadipose plate (143: 1; Fig. 27B ). The main difference between the two genera is that Spectracanthicus has lost evertible cheek plates and the modified opercle diagnostic of the Ancistrini . Oligancistrus appears to be in the process of losing evertibility of the cheek plates (they are only weakly evertible when compared to those of other members of the Ancistrini ) and the opercle is intermediate between that of closely related Ancistrini and Spectracanthicus .

Given the many reversals associated with the cheek, Spectracanthicus is a very well-diagnosed genus as it now stands. Oligancistrus , however, is not. Given the strong support for Spectracanthicus + Oligancistrus as monophyletic ( Oligancistrus shows trends towards losing the cheek armature and is not-well diagnosed; the two genera are currently monotypic and monotypic genera cannot express phylogeny), Oligancistrus is recognized as a synonym of Spectracanthicus .

Pseudancistrus has traditionally been identified by the presence of hypertrophied odontodes along the snout in both males and females and by an inability to evert the cheek plates. Among other Ancistrini , only Pseudolithoxus , Lithoxancistrus , and some members of Guyancistrus share the presence of hypertrophied snout odontodes in males and females with Pseudancistrus . It is clearly not closely related to Pseudolithoxus , although it shares numerous synapomorphies with Lithoxancistrus and Guyancistrus. A monophyletic group consisting of Hemiancistrus megacephalus, Guyancistrus , Lithoxancistrus , and Pseudancistrus is well supported with a decay index value of five and the following synapomorphies: loss of a suture between the pterotic-supracleithrum and hyomandibula (34: 0), loss of contact between the hyomandibula and prootic (35: 1), a spoon-shaped anterior process of the metapterygoid (58: 1), a thin nasal bone (105: 0), a sphenotic that does not contact the posteriormost infraorbital (117: 1), and a short ventral ridge of the basipterygium (172: 1). To rectify the paraphyly and retain Guyancistrus and Lithoxancistrus as valid genera, it is likely that several more poorly diagnosed genera would have to be described; thus, the best solution is to place Guyancistrus and Lithoxancistrus in the synonymy of Pseudancistrus and to transfer H. megacephalus to Pseudancistrus .

Rapp Py-Daniel (1985) describes Dekeyseria for two species ( D. amazona and D. scaphirhyncha ). In addition, she suggests that Peckoltia brachyura , P. picta , and P. pulcher are also Dekeyseria , but does not formally place the species in Dekeyseria . Schaefer (1986) does place the species in Dekeyseria , but Burgess (1989) and Burgess & Finley (1996) retain them in Peckoltia . All the species mentioned are unique among the Ancistrini for a combination of the presence of highly keeled lateral plates (198: 1), three rows of plates along the caudal peduncle, and several additional synapomorphies: reversal to posteriorly placed interhyal (26: 0; Fig. 13B ), an enlarged neural arch anterior to the first dorsal-fin pterygiophore (125: 1), a reversal to thin ribs (129: 0), a trapezoidal cleithrum (155: 1), and a straight anterolateral process of the basipterygium (167: 2; Fig. 33B ). Schaefer (1986) is correct in placing the species in Dekeyseria ( Fig. 38 ). In addition, Plecostomus niveatus La Monte has highly keeled lateral plates and three rows of plates on the caudal peduncle (pers. observ.), and is recognized here as Dekeyseria niveata . Isbrücker et al. (2001) recognize Zonancistrus for some species of Dekeyseria based on coloration (alternating brown and tan bands in Zonancistrus and grey in Dekeyseria ). However, there is no reason to recognize Zonancistrus as distinct based simply on colour differences, and I recognize it as a synonym of Dekeyseria .

The relationships of the genera of the Ancistrini in Schaefer’s (1986) study and this study ( Fig. 38 ) differ in many respects. The greatest similarities are the recognition of Chaetostoma , Dolichancistrus , and Leptoancistrus as a clade and this group plus Ancistrus , Exastilithoxus , Lasiancistrus , and Lithoxus as a clade. In this study, Lasiancistrus s.s. is found to be the sister of Ancistrus instead of Chaetostoma as determined by Schaefer (1986). Support for a clade of Ancistrus + Lasiancistrus comes mainly from the recognition of the presence of snout tentacles/tentacules in Lasiancistrus (208: 1/2). Ancistrus has long been diagnosed by the presence of elongate fleshy structures (tentacles, 208: 2) on the top of the snout in males. Although Lasiancistrus does not have tentacles as long as those in Ancistrus , short tentacules are present on the snout plates (208: 1; Sabaj et al., 1999). Several other characteristics corroborate the evolution of these snout tentacles: a spindle-shaped hypohyal (21: 1), a slender quadrate (64: 0) and tentacules on the pectoral-fin spine that are larger than their supporting odontodes (209: 2).

The placement of Lasiancistrus as sister to Chaetostoma by Schaefer (1986) is based on three putative synapomorphies. The first is an extension of the quadrate for articulation with the canal plate (65: 1; Fig. 13B, D ) which is found in this study to have evolved independently in several lineages. The second is a sculpturing of the anterior edge of the anterohyal; however, the states seen in Lasiancistrus and Chaetostoma are not homologous (see Fig. 8B vs. D). The last, a mesial process on the second branchiostegal (6: 1), is not present in the Lasiancistrus I examined.

Another major difference between this study and that of Schaefer (1986) is the placement of Pseudacanthicus . In this study it is part of a large clade with the rest of the Acanthicus group, Hypancistrus, Parancistrus , Panaque , and Peckoltia (referred to below as the Panaque clade). In all of the Panaque clade except Acanthicus and Peckoltia oligospila , the dentaries form an angle averaging less than or equal to 80∞ (69: 1) and the mesethmoid disk extends anterior to the main body of the mesethmoid (101: 1). In addition, all except P. oligospila have a longitudinal ridge on the quadrate (68: 1; Fig. 15H, I ). Support in Schaefer (1986) for the placement of Pseudacanthicus with Hemiancistrus (the latter is a combination of several unrelated taxa in his analysis), Ancistrus , the Lithoxus group, Lasiancistrus , and Chaetostoma is based on the attachment of the canal plate to the suspensorium (which I found in nearly all the Ancistrini ) and a ridge (or process) ventrally on the suspensorium for the attachment of the canal plate. The process contacted by the canal plate is on the preopercle in Pseudacanthicus while it is on the quadrate in the other taxa (when present), and I did not consider the two processes to be homologous.

The only described genus of the Ancistrini that was not examined for this study is Hopliancistrus Isbrücker and Nijssen. It is difficult to speculate on the phylogenetic position of Hopliancistrus based on the specimens I have examined. It appears to be very similar to Lasiancistrus , but lacks whiskerlike odontodes. It shares with Ancistrus and Lasiancistrus the presence of very strongly evertible cheek plates and very strong hypertrophied odontodes associated with them, and probably belongs along the branch that includes Ancistrus and Lasiancistrus ; however, Hopliancistrus must be examined in detail before its relationships can be determined.

Figures

Figure 8. Hypohyal, anterohyal, and posterohyal bones, right side, ventral view. A, Astroblepus sp., USNM 302674. B, Chaetostoma sovichthys, INHS 34957. C, Hypostomus unicolor, USNM 319355. D, Lasiancistrus maracaiboensis, INHS 60465. E, Leporacanthicus galaxias Isbrücker and Nijssen, INHS 40910. Scale bars = 1 mm. Shaded area is cartilage.

Figure 13. Suspensorium, right side, mesial view. A, Cordylancistrus torbesensis, MCNG 8066. B, Isbrueckerichthys duseni, UMMZ 215262. C, Lasiancistrus maracaiboensis, INHS 60465. D, Neblinichthys pilosus Ferraris, Isbrücker, and Nijssen, AMNH 56138SW. Scale bars = 1 mm. Shaded area is cartilage.

Figure 15. Suspensorium, right side, lateral view. A, Astroblepus sp., USNM 302674. B, Dolichancistrus cobrensis (Schultz), MCNG 6470. C, Delturus anguilicauda (Steindachner), USNM 318209. D, Hemiancistrus sp. 1, UF 77850. E, Hypostomus unicolor, FMNH 101120. F, Isbrueckerichthys duseni, UMMZ 215212. G, Lithoxus lithoides, BMNH 1972.7.17: 66–115. H, Pseudacanthicus leopardus (Fowler), FMNH 95554. I. Spectracanthicus murinus, MZUSP 34279. Scale bars = 1 mm. Shaded area is cartilage. In B, the lateral wall of the pterygoid channel is deflected at a right angle so that it appears as a ridge in the drawing when it is as tall as the mesial wall. Arrows point to highly deflected preoperculohyomandibular crests in D and I and to a furrow in the metapterygoid that is the precursor to the pterygoid channel in C.

Figure 20. Lateral ethmoid, right side, ventral view. A, Chaetostoma pearsei, INHS 345889. B, Delturus anguilicauda, USNM 318209. C, Hemipsilichthys sp., USNM 320377. D, Hypostomus unicolor, USNM 319355. Scale bars = 1 mm. Shaded area is cartilage. Dashed line indicates the extent of the nasal capsule.

Figure 27. Adipose fin and preadipose plates, left side, lateral view. A, Hypostomus panamensis, INHS 36095. B, Spectracanthicus murinus, MZUSP 34279. C, Delturus parahybae Eigenmann and Eigenmann, FMNH 59734. Scale bars = 1 mm.

Figure 33. Pelvic basipterygium, ventral view. A, Astroblepus sp., USNM 318209. B, Delturus anguilicauda, USNM 318209. C, Hypostomus plecostomus, ZMA 105.306. D, Kronichthys sp., MZUSP 27545. E, Peckoltia ucayalensis, INHS 40916. F, Pseudorinelepis genibarbis, INHS 39730. Scale bars = 5 mm.

Figure 35. Inner dentary teeth, right side, ventral view. A, Hypostomus plecostomus 3, INHS 28903. B, Hypostomus hemicochliodon, FMNH 97010. C, Hypostomus plecostomoides (H. cochliodon group), INHS 59831. D, Scobinancistrus pariolispos Isbrücker and Nijssen, ZMA uncatalogued. Scale bars = 0.5 mm.

Figure 38. Relationships of the taxa within the Ancistrini based on the ordered analysis. This is part of the strict consensus of 5098 most parsimonious trees of trees of 1328 steps, CI = 0.203, remainder of the strict consensus tree is in Figs 36 and 37. Numbers above the branches are clade numbers, numbers below are decay indices. 1Hemiancistrus landoni represents an undescribed genus. 2Oligancistrus is placed in Spectracanthicus. 3Lithoxancistrus is transferred to Pseudancistrus. 4Cordylancistrus platyrhynchus is transferred to Chaetostoma.