Redescription of Panaqolus purusiensis (LaMonte, 1935) (Siluriformes: Loricariidae) with identification key to the species of the genus

Despite Panaqolus purusiensis being described nearly 80 years ago, very little is known about it. The taxon was described based on a single specimen. Researching collection catalogues has revealed two more specimens that were caught together along with the holotype but had subsequently been deposited in different museums. Recent collections, including one from the type locality, have made more specimens available. Examination of this new material reveals that this species has three distinct color patterns that are size dependent. Specimens with standard length (SL) less than 30 mm show a coloration common among various species of Panaqolus consisting of regular bands on the body, fins and head. Specimens from 30-90 mm SL have a specific color pattern with thinner bands and those larger than 90 mm SL have a uniformly dark body with only the fins continuing to show bands. The most similar species are P. changae, P. gnomus, P. maccus, and P. nocturnus. The first three can easily be distinguished by their coloration and by body proportions. Specimens of P. nocturnus of more than 90 mm SL however are nearly indistinguishable from P. purusiensis of the same size, making identification difficult.


Introduction
published a revision of the small Panaque and defined a Panaque dentex species group.In 2001, Isbrücker & Schraml described the new genus Panaqolus for this group, with P. gnomus as type species.Initially this new genus was treated as a synonym of Panaque (Chockley & Armbruster, 2002;Fisch-Muller, 2003) or as a subgenus (Armbruster, 2004(Armbruster, , 2008;;Lujan et al., 2010).But later in their checklists, Ferraris (2007) and Eschmeyer (2012) considered Panaqolus as a valid taxon.Based on molecular data, Cramer et al. (2011) found Panaqolus to be clearly distinct from Panaque, but included only a few representatives of Hypostominae as the focus of the study was on two other loricariid subfamilies.Recently, Fisch-Muller et al. (2012) described a new species of Panaqolus, thus again recognizing the genus.
Molecular data show Panaqolus to be a distinct genus, with Peckoltia, Hypancistrus, and Scobinancistrus being the most closely related taxa (Cramer et al., 2011).Fisch-Muller et al. (2012) found Panaqolus and Peckoltia to be closely related analyzing cytochrome c oxidase I (COI) sequences from Hemiancistrus, Panaqolus, Peckoltia, and Pseudacanthicus.Panaqolus lacks a posterior orbital notch and a ventrolateral keel on the caudal peduncle (characters present in Panaque).Furthermore, Panaqolus remain smaller (<15 cm SL vs. >27.8cm with some species of Panaque reaching more than 60 cm SL).Panaqolus and Panaque both possess acutely angled rows of robust spoon-shaped teeth as a common character that is additionally shared with the Hypostomus-cochliodongroup.The genus Panaqolus is accepted and used for the species redescribed herein, however, a phylogenetic study is in preparation to analyze these relationships in more detail.
The description of Panaqolus purusiensis was based on a single specimen collected by B. A. Krukoff in 1934.Apparently, LaMonte did not know that two more specimens had been collected together with the holotype, but had been deposited in different collections (Conci & Michelangeli, 1974;Schaefer & Stewart, 1993).Only one of these additional specimens (USNM 94665) had been examined by Schaefer & Stewart (1993) in their revision.LaMonte also had not been aware that there had already been an available description of another small species of Panaque (P.dentex Günther, 1868).She, at least, does not refer to this species and her characters only distinguish P. purusiensis from large Panaque.As the species description is rather brief and does not provide many useful characters, a redescription is provided here.This is based on all three specimens collected by B. A. Krukoff, additional specimens recently collected at the type locality, two specimens from Acre State, Brazil, and three from the Purus drainage in Peru.

Material and Methods
Counts, measurements and terminology follow Lujan et al. (2010).All measurements were made to the nearest 0.1 mm with the use of digital or analog calipers.The following abbreviations were used: SL, standard length; HL, head length; and DW, oral disk width.Institutional abbreviations are as listed in Ferraris (2007)  Panaqolus purusiensis can be distinguished from P. albomaculatus by its body and fin coloration (alternating dark and light bars or uniformly brown body [in large specimens] vs. small, white or yellowish spots on body and fins [n = 26]), by the position of the dentaries (touching in the middle forming a V with approximately 70° angle vs. normally not touching, forming a very acute angle or being nearly parallel [Fig.1]), by the mouth width (42-50% HL vs. 32-44%) and the longer maxillary barbel length (10-18% HL vs. 2-10%).As P. dentex do not pass a size of 74 mm SL, a size where P. purusiensis still shows its banded pattern on body and fins, these two species can easily be separated as the former species never shows bands on body or fins (n = 20).
The lines on the head and trunk may be faded in large specimens (> 85 mm SL) of P. purusiensis.In this case, these specimens are especially difficult to distinguish from large P. nocturnus (Fig. 5).Panaqolus purusiensis usually has a higher caudal peduncle than P. nocturnus (11)(12)(13)(14)), large individuals (> 85 mm SL) have a lower thorax length to pelvic-dorsal depth ratio (86-98% SL [mean 93%] vs. 86-125% SL [mean 103%), and a lower dorsal-pectoral depth to dorsal-pelvic depth ratio (107-116% SL [mean 111%] vs. 110-142% SL [mean 122%]).Panaqolus koko (n = 1) has uniformly dark body and fins (vs.banded) and nearly quadrate teeth with strong lateral cusps (vs.round teeth without or with small lateral cusps).1. Medium-sized loricariid with standard length of measured specimens up to 130.4 mm SL.Dorsal profile of head and snout strongly convex from snout tip to posterior tip of supraoccipital, straight and posteroventrally slanted between dorsal-fin origin and adipose-fin origin, gently concave through caudal peduncle to posterior tip of last procurrent caudal fin ray.Dorsal orbit margin slightly raised, forming a gentle ridge, narrowing anteriorly, from anterior orbit margin to area lateral to nares.Dorsal surface of trunk transversely flattened from dorsal-fin origin to adipose-fin base.Ventral profile of head and body flat from oral disk to anal-fin origin.Caudal peduncle oval in cross-section and relatively deep (11-14% SL).

Description. Counts and proportional measurements in Table
Greatest body depth at dorsal-fin origin.Pectoral-fin origin just posterior to orbit; pelvic-fin origin at vertical through origin of second dorsal-fin ray; anal-fin origin shortly posterior after vertical through origin of last dorsal-fin ray.Adipose fin with well-ossified leading spine bearing odontodes.
Dorsal fin II,7-8 (holotype II,8), pectoral fin I,6, pelvic fin i,5, anal fin i,4, caudal fin i,13-14,i (holotype i,13,i).Spinelet V-shaped, dorsal-fin spine lock functional, posterior fin margin straight, margin of last two rays rounded.Dorsal-fin origin closer to vertical line passing through pelvic-fin origin than to vertical line passing through pectoral fin origin; not reaching adipose fin when adpressed.Adipose fin triangular; adipose-fin spine slanted posteroventrally, tip straight to curved ventrally, pointed; posterior margin of adipose-fin membrane concave to nearly vertical.Pectoral-fin spine robust, membrane of interspace between spine and first ray may bare fleshy extension in large individuals, posterior fin margin straight, when depressed reaching 1 / 3 of pelvic fin.Pelvic-fin spine robust, posterior margin slightly rounded, when adpressed reaching mid-length of anal fin.Caudal fin forked.
Head and body entirely plated except for small naked area around dorsal-fin base, snout without naked area near tip.Abdomen of adults ranging from incompletely plated (plates only in skin ventral to pectoral girdle, along sides, and in skin posterior to pelvic fin) to completely covered by small irregularly arranged platelets (specimens > 85 mm SL).Abdomen of juveniles unplated.Area dorsal to pelvic-fin base below ventral margin of inframedian plate row usually with 1-3 small plates, without smaller platelets and with large intervening naked space.Supraoccipital bordered posteriorly by 2-3 (mode 2) scutes on each side.Body with pronounced lateral ridge extending from cleithrum to posterior margin of fifth or sixth plate of the inframedian plate row; ridge gradually decreasing in prominence posteriorly.Trunk without elevated ridges.
Head and body covered by odontodes of uniform size and distribution.Enlarged odontodes on anterodorsal border of pectoral-fin spine.Cheek odontodes hypertrophied with tips recurved laterally, longest odontode extending to posterior cleithrum margin.Interorbital space straight or slightly convex.Eye moderately large, dorsolaterally placed; orbit diameter 14-22% HL.Iris diverticulum present.Nares small and ovoid, slightly longer than wide.
Oral disk round, distal margin of upper lip well separated from maxillary barbel base.Maxillary barbels of moderate length (10-18% HL).Lips moderately rugose, small patch of elongate fleshy papillae medial to each tooth row.Border of lips smooth.
Teeth spoon-shaped and normally unicuspid; sometimes a very small lateral cusp is present that may be reduced to a lateral dent (Fig. 6).Premaxillary teeth 5-7 per ramus (mode 5), mandibular teeth 4-7 per ramus (mode 6).Premaxillae angled at approximately 90°, dentaries at approximately 70° (Fig. 1).species from this genus.Medium sized specimens show a unique color pattern, but in large fish the bands on the trunk are fading, making them very difficult to be distinguished from P. nocturnus.Unfortunately, none of the characters mentioned by LaMonte (1935) actually serves to identify the species.She wrote: "This species differs from others described in the number of scutes bordering the supraoccipital; in its depth; the fact that the temporal plates are not carinate; its dorsal count, the length of the interopercular spines; and the tooth count which is lower than that of the other species."More than likely she only referred to the large species of Panaque and did not know about Panaqolus dentex, the only species of that genus which had been described at that time.The number of scutes bordering the supraoccipital varies from two to three, a number also found in congeners (vs.apparently only one in Panaque)."Its depth" probably refers to the body depth, Distribution and Habitat.The species is known from the upper Purus drainage (Fig. 7).Some of the specimens were caught at night with a cast net on stony bottom in strong current (MCP 45733).Others had been picked by hand from holes in rotten wood in low current (UFRO-I 17720).

Discussion
Panaqolus purusiensis was described some 80 years ago, but still very little is known about it.While searching museum databases for specimens for this project, individuals identified as P. purusiensis where found from localities in Peru, Brazil, Colombia and even French Guiana.All these specimens had been collected before the publication of the revision of the Panaque-dentex-group (Schaefer & Stewart, 1993).The only other species of small Panaque (now Panaqolus) that was described before 1993 is P. dentex (Günther, 1868), although it had been published as Chaetostomus dentex.That might be the reason why P. purusiensis is the most commonly used name for specimens collected before the revision of Schaefer & Stewart (1993), without regard to where had been caught.
It is not clear why and when the material collected by Krukoff in 1934 was split and deposited in at least three different collections on two continents (AMNH, New York; USNM, Washington D.C.; MSNM, Milan).LaMonte apparently did not know about this additional material, but for the present study, all three specimens of Panaqolus collected by Krukoff could be examined.Unfortunately, these three have lost virtually all coloration.Specimens from more recent collections show that very small individuals have a color pattern very similar to other  albeit P. purusiensis do not differ from other Panaqolus on this point; Panaque subjectively have a higher body, but the measurements taken here do not support this statistically.The temporal plates of all Panaqolus are not carinate, but those of at least some Panaque species are.The dorsal fin of the holotype has eight rays, but all other examined specimens have seven, as usual in Ancistrini.Among 150 examined Panaqolus only one more individual was found to have eight dorsal rays (P.nocturnus; AUM 45558).The same findings were made for Pseudancistrus barbatus (one out of 174) and Parancistrus aurantiacus (one out of 15), showing that these variations are rare but normal.The cheek odontodes of P. purusiensis are not longer than in other congeners, but Panaque may have remarkably long odontodes (reaching past the second plate after pectoral-fin insertion).The tooth count in P. purusiensis is more or less the same as in other species of the genus, but less than the number in Panaque (3-8 per ramus vs. 4-24 [Lujan et al., 2010]).
Schaefer & Stewart (1993) mention that P. purusiensis and P. nocturnus could be distinguished by a smaller orbit diameter, a deeper caudal peduncle, and less scutes in the lateral line.Unfortunately these findings could not be confirmed.The eyes of both species have nearly the same size (13.6-21.8%HL vs. 14.2-20.0%HL), the caudal peduncle usually is a little bit deeper, but with a large overlap (10.9-14.0%SL [mean 12.1%] vs. 9.8-12.1% SL [mean 11.1%]), and P. purusiensis has 24-25 [mode 24] plates in the lateral line versus 23-26 [mode 25] in P. nocturnus, showing again a tendency but no clear distinction.Despite these similarities, there is no doubt that P. purusiensis and P. nocturnus are distinct species as their distributions are clearly separated (rio Purus basin vs. rio Marañón basin; Fig. 7) and the trunk coloration of juvenile and subadult specimens is noticeably different (alternating light and dark bands vs. diffuse lighter markings on dark ground, but never showing bands).
The genus Panaqolus is not widely accepted, but previous results from molecular studies showed the genus to be clearly separated from Panaque (Cramer et al., 2011;Nathan Lujan, unpublished data, pers. comm.).Schaefer & Stewart (1993) list some morphological characters to separate the two genera, but Armbruster (2004) found them to be closely related and treated them as synonyms.Given the morphological characters from Schaefer & Stewart (1993) and first molecular results, the genus Panaqolus is being used here, following other recent publications (Eschmeyer, 2012;Ferraris, 2007;
with the addition of UFRO-I for Ichthyological Collection of Universidade Federal de Rondônia, Brazil.
body and fins (dark bars wider than light interspaces vs. dark bars as wide as or thinner than the lighter interspaces) and the larger adult size (130 mm SL vs. 86 mm SL).Panaqolus purusiensis differs from P. gnomus by the posterodorsal to anteroventral orientation of bars on the trunk (vs.anterodorsal to posteroventral [n = 30]) and the larger adult size (130 mm SL vs. 70.8mm SL).From P. maccus it can

Table 1 .
Lujan et al., 2010)ic features of Panaqolus purusiensis and P. nocturnus.Values are given as percent of standard length or head length.SD = standard deviation, n = number of specimens, H = holotype.Interlandmarks (ILM) are the two points between which measurements were taken (fromLujan et al., 2010).