New species of the Pseudancistrus barbatus group (Siluriformes, Loricariidae) with comments on its biogeography and dispersal routes

Abstract A new species of Pseudancistrus is described from the Tapajós Basin, and assigned to the P. barbatus group by having hypertrophied odontodes along the snout and lacking evertible cheek plates. The new species is distinguished from other species in that group (P. barbatus, P. corantijniensis, P. depressus and P. nigrescens) by its pattern of spots, length and color of snout odontodes, greater head depth, cleithral width, anal-fin spine length, peduncle depth and internares width. Molecular phylogenetic results corroborate placement of the new species in the Pseudancistrus barbatus group which is otherwise distributed in the Xingu Basin and rivers draining the Guyana Shield into the Atlantic Ocean. Topology tests strongly reject alternative hypotheses supporting close relationships with Guyanancistrus, Lithoxancistrus or the species Pseudancistrus pectegenitor, P. sidereus and P. genisetiger. Additionally, we propose two hypotheses on the distribution of the new species in the rio Tapajós, a Brazilian Shield drainage. The first one proposes that ancestral stock of the P. barbatus group was widely distributed throughout rivers draining the Guyana and Brazilian shields, and the species P. zawadzkii and Pseudancistrus sp. L17 are in the limit of the distribution for the group in Tapajós and Xingu rivers. The second hypothesis proposes that ancestral stock of the P. barbatus group was restricted to Guyana Shield rivers, and that headwater capture events permitted several dispersal routs through Guyana and Amazon rivers, permitted that the ancestral lineages of Pseudancistrus sp. L17 and P. zawadzkii reached the rivers of Amazon basin.


Introduction
Ancistrini is a highly diverse tribe of the subfamily Hypostominae, with 30 genera (Lujan and Armbruster 2011;Covain and Fisch-Muller 2012;Salcedo 2013) and 252 valid species (Eschmeyer and Fong 2013) widely distributed in the Neotropics from rivers in Panamá to the La Plata system in Argentina. Armbruster (2004a) provided morphological support for the monophyly of Ancistrini based on his extensive analysis of relationships within Loricariidae. Molecular data, however, suggested that Ancistrini is not monophyletic (Montoya-Burgos 1998; Covain and Fish-Muller 2012).
In this paper, we present a formal description of a new species of Pseudancistrus from the Tapajós river basin. Additionally, we provide a phylogenetic context for the new species based on analysis of sequence data of F-reticulon 4 nuclear gene, and a brief discussion of biogeographic scenarios that may explain the distribution of the new species in the rio Tapajós and northern Brazilian Shield.

Sampling and morphological analysis
After capture, fish were anesthetized using 1% benzocaine in water, and either preserved in 95% ethanol for molecular studies or fixed in 10% formaldehyde for morphological studies. Vouchers and tissues were deposited in the collection of the Laboratório de Biologia e Genética de Peixes (LBP) and Museu de Zoologia da Universidade de São Paulo (MZUSP), Brazil, Muséum d'histoire naturelle de la ville de Genève (MHNG), Switzerland, Academy of Natural Sciences of Philadelphia (ANSP) and Auburn University (AUM), U.S.A., and Smithsonian Tropical Research Institute (STRI), Panama. Measurements and counts were taken on left side of specimens. Measurements follow Armbruster (2003), and were taken point to point to the nearest 0.1 mm with digital calipers.
The products were then identified on a 1% agarose gel. The PCR products were purified using ExoSap-IT® (USB, Affymetrix Corporation, Cleveland, Ohio) following the manufacturer's instructions. The purified PCR products were used to make a sequencing PCR using the BigDyeTM Terminator v 3.1 Cycle Sequencing Ready Reaction Kit (Applied Biosystems-Life Technologies do Brasil Ltda, Vila Guarani, SP, Brazil). Subsequently, the amplified DNA was purified again and loaded onto a 3130-Genetic Analyzer automatic sequencer (Applied Biosystems), in the Instituto de Biociências, Universidade Estadual Paulista, Botucatu, São Paulo. Contigs were as-sembled and edited in BioEdit 7.0.9.0 (Hall 1999). Where uncertainty of nucleotide identity was detected, IUPAC ambiguity codes were applied. All sequences obtained in this study were deposited in GenBank (Table 3).

Sequence alignment and phylogenetic analyses
The DNA sequences were aligned using ClustalW program implemented in DAMBE 5.2.31 (Xia and Xie 2001) and edited in BioEdit 7.0.1 (Hall 1999), using default parameters. The alignments were inspected by eye for any obvious misalignments that were then corrected. Alignment errors only were changed where indels of 1 bp were added to introns of the reticulon gene. The sequence of F-reticulon 4 of the new species was sequenced twice, and a preliminary phylogenetic analysis was performed to control potential sequencing errors involving pseudogenes, paralogous copies or laboratory crosscontamination or mistakes during manipulations of samples. Nucleotide variation was examined using MEGA 5.0 (Tamura et al. 2007). To evaluate the occurrence of substitution saturation, we estimated the index of substitution saturation (Iss) in DAMBE 5.2.31 (Xia and Xie 2001), as described by Xia et al. (2003) and Xia and Lemey (2009).
Maximum-Likelihood (ML) analyses were performed using RAxML Web-Servers (Randomized Accelerated Maximum Likelihood, Stamatakis et al. 2008) which implements a faster algorithm of heuristic search with bootstrap pseudoreplicates (RBS). Bootstrap resampling (Felsenstein 1985) was applied to assess support for individual nodes using 1,000 replicates. Random starting trees were used for each independent ML tree search and all other parameters were set on default values. The ML analysis was conducted under a Generalized Time Reversible (GTR) model, with Gamma distribution (G) and Invariable Sites according to Modeltest 3.7 results (Posada and Crandall 1998). Gaps were treated as missing data.
Alternative tree topologies were evaluated in the program Treefinder (Jobb et al. 2004) using the Shimodaira and Hasegawa (SH) test (Shimodaira and Hasegawa 1999), the Approximately Unbiased (AU) test (Shimodaira 2002), and the Expected Likelihood Weights (ELW) method (Strimmer and Rambaut 2002). All tests were conducted under ML with a GTR model and Gamma distribution.
Diagnosis. Pseudancistrus zawadzkii is distinguished from all congeners, except species of the P. barbatus group, by presence of hypertrophied odontodes along the snout margin and the lack of evertible cheek plates. It further differs from two members of that group, P. barbatus and P. depressus, by having whitish spots that abruptly increase in size between the head (diameter 1.1−1.3 mm) and body (diameter 2.6−3.0 mm) (vs. whitish spots very small on whole body less than 1 mm), and snout odontodes yellowish (vs. snout odontodes reddish-brown). The new species differs from the other two members of the P. barbatus group, P. corantijniensis and P. nigrescens, by having odontodes along margin of snout increasing gradually in length from posterior of snout tip to cheek (vs. length of snout odontodes more uniform, smaller on tip of snout) and by having odontodes relatively longer on the most posterior portion of the nonevertible check plates (  Burgos 2008 for comparison of both characters). Additionally, P. zawadzkii differs from P. nigrescens by having rounded spots that do not cover more than one plate along the body (vs. whitish spots that become hazier along the body and can cover more than one plate, see P. nigrescens in fig. 3  ; shorter distance between posteromedial margin of supraoccipital and origin of dorsal-fin, 6.7−9.2% of SL (vs. 10.4−11.6% of SL in P. nigrescens); greater anal-fin spine length, 11.9−13.8% of SL (vs. 7.3−10.4 of SL in P. barbatus); greater peduncle depth, 12.5−14.2% of SL (vs. 9.3−10.4 of SL in P. barbatus); and wider internares distance, 12.7−16.6% of HL (vs. 9.9−11.8% of HL in P. barbatus). Pseudancistrus zawadzkii differs from P. kwinti and P. guentheri, two probable members of P. barbatus group by having whitish spots of the body (vs. body mottled or with bars, in P. kwinti and body plates dark at the base and pale along the edges, in P. guentheri).
Description. Morphometric data presented in Table 1. In lateral view, dorsal profile convex from snout tip to dorsal-fin origin; straight, gradually descending from dorsal-fin origin to posterior insertion of adipose fin; straight, steeply ascending to insertion of caudal fin; ventral profile flat from snout tip to anal-fin origin; shallowly concave from anal-fin insertion to lower caudal-fin spine; greatest body depth at dorsal-fin origin. In dorsal view, greatest body width across cleithral region; snout broadly elliptical; body progressively narrowed from opercular region to caudal fin. Cross-section of body between pectoral and pelvic fins rounded dorsally and flattened ventrally; cross-section of caudal peduncle ellipsoid.
Body plates and cleithrum have minute odontodes. Odontodes slightly hypertrophied on pectoral-fin spines, becoming gradually larger towards tips. Numerous yellowish hypertrophied odontodes along lateral margins of head including snout; odontodes small on tip of snout, increasing gradually in length from anterolateral margin of snout to cheeks; longest odontodes on posterior most portion of non-evertible cheek plates. Eyes small (orbital diameter 14.5−18.8% of HL), dorsolaterally positioned. Oral disk transversely ellipsoid. Lower lip not reaching transverse line between gill openings. Lower lip covered with numerous small papillae. Maxillary barbel developed. Mouth relatively large. Premaxillary teeth 40−61 per ramus; dentary teeth 28−69 per ramus. Teeth bifid, medial cusp large and rounded, lateral cusp minute and pointed. Wide jaws, dentary bones forming an oblique angle, premaxillary bones almost co-linear.
Dorsal fin II,7, origin approximately at midpoint between pectoral-and pelvic-fin origins, last dorsal-fin ray reaching adipose fin when depressed. Pectoral fin I,6, spine tip curved inward, covered with enlarged odontodes distally; depressed tip reaching one-third length of pelvic-fin spine. Pelvic fin I,5, spine tip curved inward, almost reaching anal-fin origin when depressed. Anal fin I,5, spine tip straight, reaching sev-enth plate posterior to its origin. Caudal fin I,7−I,7, distal margin concave, inferior lobe longer than superior. Adipose fin with lightly curved spine, preceded by single median preadipose plate.
Color in life. Ground color dark greenish-brown on dorsum and sides of body, becoming dark brown posteriorly, and lighter brown ventrally. Anterior portion of head to posterior margin of orbits with many small, crowded, yellow spots; spots becoming abruptly larger on posterior portion of head, continuing on body, becoming slightly and gradually larger towards caudal peduncle. Dorsal plate series usually with two large spots per plate. Mid-dorsal plates usually with one large spot per plate. Lateral median plates with one large spot per plate. Mid-ventral plates and ventral plates with one large spot per plate. Dorsal-fin spine, rays and membranes with large round large spots. Adipose-fin with two large spots on spine and membrane. Pectoral, pelvic, anal and caudal fin with numerous and similarly sized yellow spots. Hypertrophied odontodes along head margin yellowish (Fig. 3).
Color in alcohol. Similar to pattern described for living individuals, but with ground color dark brown, and spots pale tan (Fig. 1).
Sexual dimorphism. Males possess a papilla posterior to urogenital opening, an attribute absent in females. Both sexes in P. zawadzkii exhibit highly hypertrophied odontodes along snout margin, similar to others species of Pseudancistrus (Armbruster 2004b). In some loricariid species of genus Pareiorhaphis those hypertrophied odontodes may be sexually dimorphic (Pereira et al. 2007), an attribute not observed in the new species P. zawadzkii.
Etymology. Specific name is in honor of Cláudio Henrique Zawadzki, professor at Universidade Estadual de Maringá (UEM), Maringá, Paraná State, Brazil, in recognition of his dedication and remarkable contributions to the study of the family Loricariidae.   Distribution. Pseudancistrus zawadzkii is known from rio Tapajós (04°33'10"S, 56°18"W) and rio Tracuá (04°28'11"S, 56°17'01"W), municipality of Itaituba, all from rio Tapajós basin, Pará State, Brazil. (see Fig. 4 for distribution map of type species localities). Ecological notes. The rio Tapajós, and rio Tracuá where P. zawadzkii occurs are clear water rivers, varying from medium to large size, with rocky outcrops forming small waterfalls and substrates of rocks and sand (Fig. 5).

Phylogenetic analysis
Partial sequences of the nuclear gene F-reticulon 4 (RTN4) were obtained in this study and from GenBank for 44 specimens representing 35 Loricariidae species and the new species P. zawadzkii (Table 3). We included samples of the four lineages of Pseudancistrus proposed by Covain and Fisch-Muller (2012) to test whether P. zawadzkii is part of the P. barbatus group. Corydoras oiapoquensis Nijssen, 1972 (Callichthyidae) was used to root the phylogeny. Additionally, samples of Delturinae (Hemipsilichthys gobio Lutken, 1874) and Loricariinae (Harttia guianensis Rapp Py- Daniel & Oliveira, 2001) were included in the analysis as additional outgroups. The combined sequence data resulted in a matrix with 2,318 base pairs (bp), out of which 1,079 were conserved and 896 were variable. The estimated index of substitution saturation (Iss) performed in DAMBE 5.2.31 (Xia and Xie 2001) showed that the data was not saturated (i.e. Iss.c value greater than Iss).
Evolutionary relationships among species of Pseudancistrus sensu lato and other members of Otothyrini are similar between our ML phylogenetic tree (-lnL = 11470.59) and the one proposed by Covain and Fisch-Muller (2012). In our analysis, the genus Pseudancistrus is paraphyletic with species assigned to three different lineages. The first lineage is monotypic, composed of P. genisetiger, sister to H. gobio, an outgroup taxon. Covain and Fisch-Muller (2012) suggested that P. genisetiger represents an undescribed genus within Delturinae. The second lineage of Pseudancistrus (P. sidereus + P. pectegenitor ) is sister to a species of Lithoxus Eigenmann, 1910;Covain and Fisch-Muller (2012) suggested that the two species represent an undescribed genus or may be included in Lithoxus. The third lineage is composed of members of the P. barbatus group (P. depressus, P. barbatus, P. corantijniensis, P. nigrescens, the new species P. zawadzkii and an undescribed species from the rio Xingu known as L17 among hobbyists). The P. barbatus group forms a polytomy with almost all species analyzed in the ingroup (Fig. 3), and was recognized by Covain and Fisch-Muller (2012) as true Pseudancistrus since this group includes the type species P. barbatus. Additionally, Covain and Fisch-Muller (2012) revalidated two genera for several species previously assigned to Pseudancistrus, − Lithoxancistrus (for Pseudancistrus orinoco (Isbrücker, Nijssen & Cala, 1988)) and Guyanancistrus (for Pseudancistrus sp., P. brevispinis (Heitmans, Nijssen & Isbrücker, 1983), P. longispinis (Heitmans, Nijssen & Isbrücker, 1983) and P. niger (Norman 1926)). Our analysis also supports the recognition and composition of those two genera.

Taxonomy and phylogenetic comparison
The new species P. zawadzkii possesses hypertrophied odontodes along the snout margin and lacks evertible cheek plates. Armbruster (2004b) identified that among Ancistrini, only Pseudolithoxus, Lithoxancistrus, and some members of Guyanancistrus and Pseudancistrus share the presence of hypertrophied odontodes along the snout in both sexes. Armbruster (2004b) also suggested that the species of Pseudancistrus that present this characteristic are derived; those species correspond to the P. barbatus group proposed by Chambrier and Montoya-Burgos (2008). Therefore, the new species described herein is a typical member of this group sensu Covain and Fisch-Muller (2012). Our phylogenetic analysis (Fig. 3) supports that hypothesis, and places the new species in a polytomy with P. corantijniensis, Pseudancistrus sp. L17 (undescribed species) and P. nigrescens, within the P. barbatus group. Our likelihood-based tests strongly rejected alternative topologies placing the new species in Lithoxancistrus, Guyanancistrus or with other species of Pseudancistrus apart from the P. barbatus group (see Table 2).
Pseudancistrus zawadzkii, P. corantijniensis, and P. nigrescens share the presence of whitish colored snout odontodes and a dark colored body covered with white spots. The new species can be easily distinguished from P. corantijniensis and P. nigrescens by having large hypertrophied odontodes on the posteriormost portion of the nonevertible check plates, and marginal odontodes that increase gradually in length from tip of snout to cheeks. Pseudancistrus barbatus and P. depressus share reddish-brown snout odontodes, a probable synapomorphy, and are the sister group to P. zawadzkii, P. corantijniensis and P. nigrescens. Covain and Fisch-Muller (2012) suggested that P. guentheri and P. kwinti may be added to the P. barbatus group. However, those two species have a different body coloration pattern see fig. 3); in P. kwinti the body is either mottled or with bars, while in P. guentheri the body plates are dark at the base and pale along the edges (Willink et al. 2010).

Biogeography and dispersal routes
Named species of the P. barbatus group are distributed in rivers draining to Guyana Shield into the Atlantic Ocean, and the new species described herein is from Tapajós river draining of Brazilian Shield into the Amazon. In our phylogeny, species from the eastern Guyana Shield (P. barbatus and P. depressus) form a clade sister to a group composed of species from the western Guyana Shield (P. corantijniensis and P. nigrescens) and Amazon basin (P. zawadzkii and Pseudancistrus sp. L17) (Fig. 6). Therefore, based on this interpretation and our results of phylogenetic analysis, we suggested two hypotheses that could generate the distribution pattern of P. barbatus group extant-species. The first hypothesis is that the ancestral stock of the P. barbatus group was widely distributed through all Guyana Shield rivers and Amazon Brazilian Shield rivers, and the species P. zawadzkii and Pseudancistrus sp. L17 are in the limit of the distribution for the group in Tapajós and Xingu rivers, respectively. Gaston (1998) and Hubbell (2001) suggested that when allopatric divergence is the dominant mode of speciation, many daughter species are expected to arise from geographically widespread ancestral species. This is a reasonable interpretation given that named species of the group are widespread in rivers draining Guyana Shield into the Atlantic Ocean; the new species P. zawadzkii are from Tapajós river drainage of Amazon Brazilian Shield; the possible new and undescribed species Pseudancistrus sp. L17 are from Xingu river which also belongs to drainages of Amazon Brazilian Shield and others possible new and undescribed species of P. barbatus group may be present in drainages of Guyana Shield into Amazon (Pseudancistrus sp. L220 from rio Paru; Pseudancistrus sp. L251 from rio Cuminá (rio Erepecuru); Pseudancistrus sp. L383 from rio Trombetas; Pseudancistrus sp. L440 from rio Jatapu (Seidel 2008)). However, phylogenetic and taxonomic studies are necessary to confirm that the latter undescribed species belong to P. barbatus group.
The second hypothesis suggests that the ancestral stock of P. barbatus group should have been distributed through Guyana Shield rivers and there existed several dispersal routes through Guyana and Amazon rivers, permitting that the ancestral lineages of Pseudancistrus sp. L17 and P. zawadzkii reached the rivers of Amazon basin (see Fig. 7 for dispersal routes). Therefore, examples of connections and areas of movement among Guyana drainages and the north tributaries of Amazon basin was reported by several authors: (1) the Rupununi portal, an example of seasonal connection among Takutu and Rupununi rivers (Armbruster and Werneke 2005;Lujan and Armbruster 2011;De Souza et al. 2012); (2) the corridor among Sipalawini (Corantijn river basin) and the Paru do Oeste (Amazon basin), also connected only in the rainy season (Nijssen 1972;Lujan and Armbruster 2011); (3) the Cassiquiare Canal, a large and permanently navigable corridor between the upper Orinoco and the upper Rio Negro (Amazon) (Chernoff et al. 1991;Buckup 1993;Schaefer and Provenzano 1993;Lovejoy and Araújo 2000;Turner et al. 2004;Moyer et al. 2005;Willis et al. 2007;Winemiller et al. 2008;Winemiller and Willis 2011); (4) Proto-Berbice, a river system which had its headwaters in an ancient  mountain range draining northward to Guyana system (Rupununi and Essequibo rivers) and suffered a major sedimentation, erosion and/or corrosion of the highlands and at the end of the Pliocene had its head waters captured by the Amazon system; (5) the Atlantic coastal corridors resulted in a coastal marine corridor with reduced salinity due to the westerly Amazon River discharge, coastal junctions during times of marine regressions Table 3. Taxa list, specimen and sequence data analyzed in the present study (n=44). Institutional acronyms follow Fricke and Eschmeyer (2013).
and expanded coastal plains, and stream captures (Eigenmann 1912;Boeseman 1968;Cardoso and Montoya-Burgos 2009;Lujan and Armbruster 2011). Additionally, the mainstream of Amazon River can act as a permeable barrier for endemic taxa on the respective Guiana and Brazilian shields. Several genera known to tolerate more lowland conditions (e.g. Ancistrus Kner, 1854, Lasiancistrus, and Hypostomus Lacepéde, 1803) may be able to cross the Amazon basin, but such dispersal is unlikely among most species of Ancistrini (Lujan and Armbruster 2011). Also historically, epochs of cooler climate, as during glacial periods, could produce reduced precipitation, marine regressions, expansion of the coastal plain, and deepening of river channels. During such arid periods, rapids would have been more widespread, and deep-channel habitats that may currently work as barriers to fish dispersal would have been reduced (Schubert et al. 1986;Latrubesse and Franzinelli 2005;Lujan and Armbruster 2011). Drier climate will hardly change the Amazon river in a rapid, but can reduce its water flow allowing fish dispersal. Among Neotropical fishes Psectro-  (1987)), Parotocinclus aripuanensis Garavello, 1988, andP. britskii Boeseman, 1974 (Loricariidae: Hypoptopomatinae) are species known to support dispersal via the northern Brazilian Shield. Also, the dispersal routes around adjacent drainages of southern and northern Guyana Shield and northern parts of the Brazilian Shield could allow the dispersal of the ancestral form of P. zawadzkii and Pseudancistrus sp. L17, as well as others ancestral species of the P. barbatus group and even species of Ancistrini (Lujan and Armbruster 2011). The movement of fish species around adjacent drainages could be explained by two hydrographic reconfiguration process: headwater capture events (geomorphological phenomenon) and marine regressions (sea level oscillation). Changes in the earth's surface involving changes in the courses of rivers, as stream captures, portions of tributaries of a river in a watershed could be "captured" by adjacent basins resulting in isolated populations and at the same time letting species to move, or disperse, between adjacent drainages (Almeida and Carneiro 1998;Bishop 1995;Wilkinson et al. 2006Wilkinson et al. , 2010Roxo et al. 2012). Montoya-Burgos (2003) hypothesized that dispersal (followed by allopatric population divergence) among Amazon and North-eastern coastal rivers probably occurred by temporary connections between adjacent rivers during periods of lower sea level about 6-5 Ma (see fig. 5 in Montoya-Burgos 2003). Cardoso and Montoya-Burgos (2009) suggested the same process to explain dispersal of P. brevispinis along coastal rivers of the Guyana. Therefore, temporary lowland connections and headwater capture events, together with the previously related hypothesis of colonization routes, likely explain the widespread distribution of the P. barbatus group extant species on Guyana and Brazilian Shields, as well as how the ancestral lineages of P. zawadzkii and Pseudancistrus sp. L17 reached the drainages of the northern Brazilian Shield, in Tapajós and Xingu rivers.