A new species of Knodus (Characiformes: Characidae), with deep genetic divergence, from the Mearim and Munim river basins, Northeastern Brazil, and evidence for hidden diversity in adjacent river basins

Rayane G. Aguiar¹, Erick C. Guimarães^1,2,3 , Pâmella S. de Brito^1,3, Jadson P. Santos³, Axel M. Katz⁴, Luiz Jorge B. da S. Dias⁵, Luis Fernando Carvalho-Costa¹ and Felipe P. Ottoni⁶

PDF: EN XML: EN | Cite this article

Abstract​

---

Introduction​

---

Knodus Eigenmann, 1911 is one of the most species-rich characid genera within the subfamily Stevardiinae (Thomaz et al.,2015; Mirande, 2019; Ferreira et al.,2021), including 35 valid species (Menezes, Marinho, 2019; Sousa et al.,2020; Fricke et al., 2022). It is distributed among the Amazon, Tocantins-Araguaia, Orinoco, Paraná-Paraguay, Parnaíba, São Francisco, and Jequitinhonha river basins (Fricke et al.,2022), with its diversity peaking in the Amazon River basin (García-Melo et al.,2019; Fricke et al.,2022).

The genus Knodus is diagnosed only by the possession of two rows of pre-maxillary teeth – the inner row with four teeth – and a scaled caudal-fin (Eigenmann, 1918; Géry, 1972, 1977). This combination traditionally used to diagnose the genus has been questioned by some authors (e.g., Schultz, 1944; Taphorn, 1992; Román-Valencia, 2000; Román-Valencia et al., 2008) who consider Knodus a synonym of the closely related genus Bryconamericus Eigenmann, 1907.

Recent phylogenetic hypotheses based on molecular and morphological data have improved our knowledge on Knodus diversity and its intrageneric relationships (e.g., Thomaz et al.,2015; Mirande, 2019; García-Melo et al.,2019). However, although some species have been described within Knodus in recent years, the alpha taxonomy of most species and related genera is still somewhat confusing (García-Melo et al.,2019; Sousa et al., 2020).

Recent sampling efforts conducted in tributaries of the Mearim and Munin river basins (Northeast Brazil) revealed the existence of a new species morphologically similar to but still distinct from Knodus victoriae (Steindachner, 1907), whose type locality is in the upper Parnaiba River basin (Steindachner, 1907). Thus, we describe a new cryptic speciesof Knodussensu Bickford et al.(2007), from the Mearim and Munim river basins, based on an integrative taxonomic approach.

Material and methods

---

Taxon sampling, specimen collection, and preservation. Specimens were captured with a manual trail-net (2 m long × 1.8 m high; mesh size, 2 mm) and euthanized in a buffered solution of ethyl-3-amino-benzoat-methanesulfonate (MS–222) at a concentration of 250 mg/L until complete cessation of opercular movements, according to animal welfare laws and guidelines (Close et al.,1996; Leary et al.,2013). Specimens selected for morphological analysis were fixed in formalin for 10 days, after which they were preserved in 70% ethanol. Molecular data were obtained from specimens fixed and preserved in absolute ethanol. Specimens for morphological analysis are included in the lists of type and comparative material. Specimens for molecular approaches are listed in Tab. 1. Type material are deposited in the following ichthyological collections: Laboratório de Biologia e Genética de Peixes, Universidade Estadual Paulista Júlio de Mesquita Filho, Botucatu (LBP); Instituto de Biologia, Universidade Federal do Rio de Janeiro, Rio de Janeiro (UFRJ); Coleção Ictiologica do Centro de Ciências Agrárias e Ambientais da Universidade Federal do Maranhão, Chapadinha (CICCAA); and Coleção Ictiológica da Universidade Estadual do Maranhão, São Luís (CIUEMA).

Morphological analysis. Measurements and counts were made according to Fink, Weitzman (1974), with exception of the scale rows below the lateral line, which were counted to the insertion of the pelvic fin. Vertical scale rows between the dorsal-fin origin and lateral line do not include the scale of the median predorsal series situated just anterior to the first dorsal fin ray. Counts of supraneurals, vertebrae, procurrent caudal fin rays, unbranched dorsal and anal fin rays, branchiostegal rays, gill-rakers, premaxillary, maxillary, and dentary teeth were taken only from cleared and stained paratypes (C&S), prepared according to Taylor, Van Dyke (1985). The four modified vertebrae which constitute the Weberian apparatus were not included in the vertebral counts and the fused PU1 + U1 was considered a single element. Osteological nomenclature follows Weitzman (1962). Comparisons with other species of Knodus were based on examined material, as well as information from literature (Zarske, Géry, 2006; Ferreira, Lima, 2006; Zarske, 2007, 2008; Ferreira, Carvajal, 2007; Román-Valencia et al., 2008; Menezes et al.,2009, 2020; Ferreira, Netto-Ferreira, 2010; Esguícero, Castro, 2014;Menezes, Marinho, 2019; Sousa et al.,2020; Deprá et al.,2021).

DNA extraction, amplification, and sequencing. DNA was extracted from fin clips using a saline buffer extraction protocol (Aljanabi, Martinez, 1997). Fragments of the mitochondrial gene cytochrome c oxidase subunit 1 (COI) were amplified with the universal primers designed by Ward et al.(2005) (FISHF1 5´-TCAACCAACCACAAAGACATTGGCAC-3´ and FISHR1 5´-TAGACTTCTGGGTGGCCAAAGAATCA-3´), primers from Melo et al. (2011) (COI L6252-Asn 5’-AAGGCG GGGAAAGCCCCGGCA G-3’ and H7271-COXI 5’-TCC TATGTAGCCGAATGGTTC TTT T-3’) and primers developed in the present study (KNODUS-TOF 5´-GGGCGATGACCAAATCTA-3´ and KNODUS-TOR 5´-AGGGTCGAAGAATGAGGTAT-3’). Polymerase chain reactions (PCR) for samples from Knodus sp. “Mearim”, Knodus cf. savannensis, Knodus sp. “Maracaçumé”, and Knodus cf. victoriae comprised a total volume of 15 µL containing 1x polymerase buffer, 1.5 mM MgCl₂, 200 µM dNTP, 0.2 µM of each primer, 1U Taq polymerase (Invitrogen), 100 ηg DNA template, and ultrapure water. The PCR cycles were as follows: 2 min at 94ºC, followed by 35 cycles of 94ºC for 30s, 54ºC for 30s and 72ºC for 1 min and 10 min at 72ºC. Polymerase chain reactions (PCR) for samples from the Knodus sp. “Munim”, and Knodus sp. “Itapecuru” comprised a total volume of 15 µl containing 1x Polymerase buffer, 400 µM dNTP, 0,4 uM of each primer, 1U Taq Polymerase (Invitrogen), 100 ηg DNA template, and ultrapure water. The PCR cycles were as follows: 5 min at 94ºC, followed by 35 cycles of 94ºC for 45s, 48°-52ºC for 45s, and 72ºC for 1 min, and 10 min at 72ºC. Amplicons were purified using ExoSAP-IT PCR Product Clean-up (Thermo Fisher Scientific) and Gel Purification Kit (GE Healthcare Systems), and sequenced using forward and reverse primers and the BigDye Terminator 3.1 Cycle Sequencing kit in a ABI 3730 DNA Analyzer (Thermo Fisher Scientific).

Data partitioning, evolutionary models, and alignment. The dataset included the COI sequences (401 base pairs, bp) of individuals from 10 Knodus species, including the species here described, specimens that we were not able to identify at the species level, as well as two Bryconamericus species. Many sequences were newly generated for this project, while others originated from the Barcode of Life Database (BOLD) and the National Center for Biotechnology Information (NCBI) (Tab. 1). Sequences were aligned using ClustalW (Chenna et al., 2003) and translated into amino acid residues using the program MEGA 7 (Kumar et al.,2016) to test if the sequences came from NUMTs (nuclear mitochondrial DNA sequences), in which case premature stop codons or indels would be expected. The best-fit evolutionary model (GTR+I+G) was selected using the Akaike Information Criterion (AIC) and the Corrected Akaike Information Criterion (AICc) by jModelTest 2.1.7 (Darriba et al.,2012), and used in all analyses, except for ABGD (Automatic Barcode Gap Discovery) which is a model-free approach based only on genetic distances.

Phylogenetic analysis. A Bayesian inference phylogenetic (BI) tree was estimated in MrBayes 3.2 (Ronquist et al., 2012) to reconstruct the evolutionary relationships among terminals using the General Time Reversible (GTR+I+G) evolutionary model. The BI analysis was conducted with the following parameters: two independent Markov chain Monte Carlo (MCMC) runs of two chains each for 10 million generations, with a tree sampling frequency at every 1,000 generations. The convergence of the MCMC chains and the proper burn-in value were assessed by evaluating the stationary phase of the chains using Tracer v. 1.6 (Rambaut et al., 2014). The final consensus tree and its posterior probabilities were generated with the remaining tree samples after removing the first 25% samples (burn-in). We used as outgroups sequences of Bryconamericus exodon Eigenmann, 1907 and B. iheringii (Boulenger, 1887). The remaining haplotypes were used as ingroups.

Species delimitation and molecular diagnoses. We implemented five distinct and independent single locus species delimitation methods based on molecular data, each of which rely on different operational criteria for species delimitation. DBC, DNA barcoding (hereafter Traditional DNA barcoding) was initially proposed by Hebert et al. (2003a,b). Since then, it has improved and gained supporters due to its practicality and efficiency (e.g., Hajibabaei et al., 2007; Coissac et al., 2016; DeSalle, Goldstein, 2019). The premise of the method consists of standardized sequencing of a specific gene for each species, followed by the organization of the sequences in virtual reference libraries. Once a species is added to the library, any individual of that species, at any ontogenetic stage, or even fragments, can be identified by simple comparison, simply by sequencing the standard gene (Hebert et al., 2003a,b). For fish and other animals, the gene used is the mitochondrial protein cytochrome c oxidase I (COI) and its effectiveness has been frequently demonstrated (e.g., Costa-Silva et al., 2018; García-Melo et al., 2019). The methodology suggests that the maximum genetic distance between individuals of the same species, based on the COI sequences, can be defined for each taxonomic group. Any difference higher than this cut-off value would represent discontinuity between species (Hebert et al., 2003a,b). The other approaches were CBB, a character-based DNA barcoding method (DeSalle et al., 2005) adapted by Ottoni et al. (2019) and Guimarães et al. (2020b); GMYC, the General Mixed Yule Coalescent method, single-threshold version (Fujisawa, Barraclough, 2013); bPTP, the Bayesian implementation of the Poisson tree processes (Zhang et al.,2013); and ABGD, Automatic Barcode Gap Discovery (Puillandre et al., 2012).

Traditional DNA barcoding (DBC). We used the Kimura-2-parameter model (K2P) (Kimura, 1980) to estimate the pairwise genetic distances between species in MEGA 7 software (Kumar et al., 2016). We considered a cutoff of 2% as sufficient to discriminate species, since this threshold is commonly inferred by species delimitations among freshwater Neotropical fish species based on COI (Jacobina et al., 2018).

Molecular diagnosis (CBB). The molecular diagnosis approach delimited the new species by the presence of a unique combination of nucleotides at particular sites. In addition, the new species was diagnosed by nucleotide substitutions following Costa et al. (2014), Ottoni et al. (2019), and Guimarães et al. (2020b). Nucleotide substitutions among lineages were optimized on the Bayesian inference topology using PAUP version 4 (Swofford, 2002). Each nucleotide substitution is represented by its relative numeric position determined through sequence alignment with the complete mitochondrial COI gene of Psalidodon paranae (Eigenmann, 1914)(KX609386.1:5503-7062), followed by the specific nucleotide substitution in parentheses. Unique nucleotide substitutions in our analysis are marked with asterisk.

General Mixed Yule Coalescent (GMYC). The GMYC is a single locus coalescent-based species delimitation approach that relies on branch lengths to establish a threshold between speciation and coalescent processes (Fujisawa, Barraclough, 2013). Here we applied the single-threshold version of the method, which usually outperforms the multiple-threshold version (Fujisawa, Barraclough, 2013). A new reduced dataset was created for this analysis using DAMBE5 (Xia, 2013), including only unique haplotypes following the requirements of this method.

The ultrametric phylogenetic tree needed for input was inferred in BEAST version 1.8.4 (Drummond et al., 2012), with the following parameters: an uncorrelated relaxed clock with lognormal distribution, a Yule Process as tree prior with 10 million generations and sampling frequency of 1,000. We used as outgroups sequences of Bryconamericus exodon and B. iheringii. The remaining haplotypes were used as ingroups. The GMYC analysis was performed on the Exelixis Lab’s server https://species.h-its.org/gmyc.

Bayesian implementation of the Poisson tree processes (bPTP). The bPTP is another single locus coalescent-based species delimitation method that differs from other similar approaches, such as GMYC, by not requiring an ultrametric tree and thus not relying on branch lengths to delimit species (Zhang et al., 2013). The method assumes that more molecular variability (number of nucleotide substitutions) is expected between haplotypes from different species than within a species (Zhang et al., 2013), establishing a threshold between speciation and coalescent processes. The reduced dataset for performing the bPTP was the same at that used in GMYC, following the requirements of this method. The input phylogenetic tree was estimated in software Mrbayes 3.2 (Ronquist et al., 2012) to reconstruct the evolutionary relationships among terminals using the General Time Reversible (GTR+I+G) evolutionary model. The BI analysis was conducted with the following parameters: two independent Markov chain Monte Carlo (MCMC) runs of two chains each for 10 million generations, with a tree sampling frequency at every 1,000 generations. The convergence of the MCMC chains and the proper burn-in value were assessed by evaluating the stationary phase of the chains using Tracer version 1.6 (Rambaut et al., 2014). The final consensus tree and its posterior probabilities were generated with the remaining tree samples after removing the first 25% samples (burn-in). The remaining haplotypes were used as ingroups. The bPTP analysis was performed on the Exelixis Lab’s web server http://species.h-its.org/ptp/, following the default parameters except for designation of a 20% burn-in. Bryconamericus exodon was chosen as the outgroup.

Automatic Barcode Gap Discovery (ABGD). The ABGD is a barcode species delimitation method that aims to establish a minimum gap that probably corresponds to the threshold between interspecific and intraspecific processes (Puillandre et al., 2012). The major advantage of ABGD when compared to the other barcode species delimitation methods is that the inference of the limit between interspecific and intraspecific processes (gap detection) is recursively applied to previously obtained groups to get finer partitions until there is no further partitioning, allowing a more refined search. Basically, the ABGD analysis indicates the number of groups (species) estimated relative to a large spectrum of p values (prior intraspecific values). A value of 0.1 indicates maximum intraspecific variability with all sequences belonging to a single species, whereas a 0.001 value indicates a small intraspecific variability with each distinct haplotype representing a different species. The reduced dataset for performing the ABGD was the same as that used in GMYC, following the requirements of this method. We ran ABGD on the ABGD server website https://bioinfo.mnhn.fr/abi/public/abgd/abgdweb.html with default parameters, except for the X value (we used an X value of 1.0 as recommended by the server for our analysis). After running the ABGD, additional molecular, morphological, or ecological characters are needed to infer the correct number of species if the analysis will follow an integrative taxonomy paradigm.

Results​

---

Knodus guajajara Aguiar, Brito, Ottoni & Guimarães, new species

urn:lsid:zoobank.org:act:4EEDDC6E-4088-4B60-828D-AF79EF84DC26

(Figs. 1–3; Tabs. 2–3)

Knodus victoriae [non Knodus victoriae (Steindachner, 1907)]. ―Guimarães et al.(2020a:90).
―Oliveira et al. (2020:6–7).

FIGURE 1 | Knodus guajajara, holotype, CICCAA 4883, 31.4 mm SL, Alto Alegre do Pindaré municipality, Igarapé Arapapá, Pindaré River drainage, Mearim River basin.

Holotype. CICCAA 4883, 31.4 mm SL, Alto Alegre do Pindaré municipality, Igarapé Arapapá, Pindaré River drainage, Mearim River basin, 03°42’26”S 46°00’25”W, Nov 2016. E. C. Guimarães & P. S. Brito.

FIGURE 2 | Knodus guajajara, CICCAA 4861, paratype, 31.9 mm SL, jaw suspensorium. A. Premaxillary. B. Maxilla. C. Dentary. Scale bar = 1 mm.

TABLE 1 | Specimens and DNA sequence information included in the study. Sequences made available by this study follow with asteristic.

Paratypes. All from Brazil, Maranhão State: CICCAA 1535, 1, 29.2 mm SL, collected with holotype. CICCAA 1585, 3, 28.0–34.7 mm SL, Alto Alegre do Pindaré municipality, igarapé Arapapá, Pindaré River drainage, Mearim River basin, 03°42’26”S 46°00’25”W, Nov 2015, E. C. Guimarães & P. S. Brito. CICCAA 4860, 2 C&S, 29.9–31.6 mm SL, Alto Alegre do Pindaré municipality, igarapé Arapapá, Pindaré River drainage, Mearim River basin, 03°42’26”S 46°00’25”W, Nov 2015, E. C. Guimarães & P. S. Brito. LBP 31041, 16, 24.5–31.4 mm SL, Alto Alegre do Pindaré municipality, Igarapé Igarapá, Pindaré River drainage, Mearim River basin, 03°45’51”S 46°08’15”W, Nov 2015, E. C. Guimarães & P. S. Brito. CICCAA 4858, 5 C&S, 22.9–27.0 mm SL, Alto Alegre do Pindaré municipality, Igarapé Igarapá, Pindaré River drainage, Mearim River basin, 03°45’51”S 46° 08’15”W, Nov 2015, E. C. Guimarães & P. S. Brito. CICCAA 1517, 2, 21.8–24.4 mm SL, Alto Alegre do Pindaré municipality, igarapé Caititu, Pindaré River drainage, Mearim River basin, 03°42’30”S 46°01’19”W, Jul 2017, E. C. Guimarães & P. S. Brito. CICCAA 4859, 2 C&S, 23.5–27.6 mm SL, Alto Alegre do Pindaré municipality, igarapé Caititu, Pindaré River drainage, Mearim River basin, 03°42’30”S 46°01’19”W, Jul 2017, E. C. Guimarães & P. S. Brito. CICCAA 1536, 2, 28.8–29.8 mm SL, Buriticupu municipality, Buritizinho River, Pindaré River drainage, Mearim River basin, 04°11’53”S 46°28’41”W, Nov 2016, E. C. Guimarães & P. S. Brito. CICCAA 4861, 1 C&S, 31.9 mm SL, Buriticupu municipality, Buritizinho River, Pindaré River drainage, Mearim River basin, 04°11’53”S 46°28’41”W, Nov 2016, E. C. Guimarães & P. S. Brito. CICCAA 1227, 1, 29.6 mm SL, Buriticupu municipality, Buritizinho River, Pindaré River drainage, Mearim River basin, 04°19’45”S 46°29’46”W, 27 Jan 2017, E. C. Guimarães & P. S. Brito. UFRJ 7023, 10, 22.8–29.3 mm SL, Alto Alegre do Pindaré municipality, igarapé Jenipapo, Pindaré River drainage, Mearim River basin, 03°51’20”S 46°11’09”W, Jul 2017, E. C. Guimarães & P. S. Brito. CICCAA 4862, 5 C&S, 23.9–29.3 mm SL, Brazil, Maranhão State, Alto Alegre do Pindaré municipality, igarapé Jenipapo, Pindaré River drainage, Mearim River basin, 03°51’20”S 46°11’09”W, Jul 2017, E. C. Guimarães & P. S. Brito. CICCAA 1321, 39, 19.4–29.2 mm SL, Miranda do Norte municipality, Mearim River basin, 04°19’45”S 46°29’46”W, Nov 2016, E. C. Guimarães & P. S. Brito. CICCAA 4863, 11 C&S, 19.7–24.4 mm SL, Miranda do Norte municipality, Mearim River basin, 04°19’45”S 46°29’46”W, Nov 2016, E. C. Guimarães & P. S. Brito. CIUEMA 1021, 8, 21.2–26.9 mm SL, Alto Alegre do Pindaré municipality, Igarapé Timbira, Pindaré River drainage, Mearim River basin, 03°32’57”S 44°39’38”W, Nov 2015, E. C. Guimarães & P. S. Brito. CICCAA 3423, 42, 19.5–30.21 mm SL, CICCAA 3424, 57, 22.35–26.9 mm SL,, Chapadinha municipality, Riacho da Raiz, RESEX Chapada Limpa, Munim River basin, 03°53’45”S 43°29’21”W, 11 Aug 2019, J. Reis, L. Oliveira, F. Ottoni, R. Fernandes & A. Silva. Chapadinha municipality, Bandeira River, Povoado Mata do Jeroca, RESEX Chapada Limpa, Munim River basin, 03°59’40”S 43°29’24”W, 11 Aug 2019, J. Reis, L. Oliveira, F. Ottoni, R. Fernandes & A. Silva. CICCAA 3425, 22, 20.22–26.71 mm SL, CICCAA 3426, 6, 28.87–32.6 mm SL, Chapadinha municipality, stream at Bairro Aldeia, Munim River basin, 03°44’53”S 43°21’32”W, 28 Jan 2019, F. Carvalho, H. Silva, L. Oliveira & I. Gôuvea. CICCAA 3467, 2, 21.22–21.43 mm SL, Anapurus municipality, stream on the road at Povoado de Paços, Munim River basin, 03°33’44”S 43°03’52”W, Sep 2019, D. Campos, J. Reis & F. Ottoni. CICCAA 4808, 3, 25.19–34.2 mm SL, CICCAA 4822, 1, 36.89 mm SL, Chapadinha municipality, stream at balneário Repouso do Guerreiro, Bairro Independência, Munim River basin, 03°44’57”S 43°20’26”W, Nov 2019, B. Furtado, M. Paiva, A. Bezerra, M. Coelho & I. Gouvêa. CICCAA 4490, 5, 24.73–29.11 mm SL, Chapadinha municipality, Riacho Fundo, Munim River basin, 03°42’20”S 43°31’46”W, 23 Sep 2018, L. Sousa, L. Oliveira & I. Gouvêa, CICCAA 4491, 3 C&S 23.26–28.19 mm SL, Chapadinha municipality, Riacho Fundo, Munim River basin, 03°42’20”S 43°31’46”W, 23 Sep 2018, L. Sousa, L. Oliveira & I. Gôuvea.

FIGURE 3 | Knodus guajajara, CICCAA 4861, paratype, 31.9 mm SL, jaw suspensorium. A. Premaxillary. B. Maxilla. C. Dentary. Scale bar = 1 mm.

TABLE 2 | Morphometric data (N = 110) of the holotype and paratypes of Knodus guajajara from the Mearim River basin. SD = Standard deviation.

Morphological diagnosis. Knodus guajajara differs from K. borki Zarske, 2008 and K. delta Géry, 1972 by having a complete lateral line (vs. incomplete lateral line) and from K. cupariensis de Sousa, Silva-Oliveira, Canto & Ribeiro, 2020 and K. geryi Lima, Britski & Machado, 2004 by having caudal fin lobes with sparse chromatophores and lacking basal blotches (vs. a dark basal blotch on each caudal fin lobe, Sousa et al., 2020; fig. 1); from Knodus borki, K. diaphanus (Cope, 1878), K. victoriae, K. heteresthes (Eigenmann, 1908), K. deuterodonoides (Eigenmann, 1914), K. longus Zarske & Géry, 2006, K. septentrionalis Géry, 1972, K. figueiredoi Esguícero & Castro, 2014, K. geryi, K. meridae Eigenmann, 1911, K. nuptialis Menezes & Marinho, 2019, K. orteguasae (Fowler, 1943), K. tiquiensis Ferreira & Lima, 2006, K. angustus Menezes, Ferreira & Netto-Ferreira, 2020, K. rufford Deprá, Ota, Vitorino-Júnior & Ferreira, 2021 and K. obolus Deprá, Ota, Vitorino-Júnior & Ferreira, 2021 by having 20–25 branched rays in the anal-fin (mode 23) (vs. 12–19, combined); and from K. tiquiensis by having a single humeral spot (vs. two). Knodus guajajara is distinguished from K. breviceps (Eigenmann, 1908) and K. savannensis Géry, 1961 by having a conspicuous round humeral blotch (vs. inconspicuous and vertically elongate); from K. dorsomaculatus Ferreira & Netto-Ferreira, 2010 by having a hyaline dorsal-fin (vs. dark blotch on the base of the first five branched dorsal fin rays); from K. alpha (Eigenmann, 1914), K. chapadae (Fowler, 1906), K. geryi, K. hypopterus (Fowler, 1943), K. mizquae (Fowler, 1943) and K. shinahota Ferreira & Carvajal, 2007 by having 4 or 5 rows of scales between the lateral line and the dorsal-fin origin (vs. 6 rows of scales); from K. cinarucoensis (Román-Valencia et al., 2008), K. gamma Géry, 1972 and K. longus Zarske & Géry, 2006 by having 12 or 13 scales in the median series between the tip of the supraoccipital spine and the dorsal-fin origin (vs. 10 or 11 rows of scales in K. gamma and 17 to 18 in K. longus); from K. jacunda (Fowler, 1913), K. moenkhausii (Eigenmann & Kennedy, 1903), Knodus cismontanus (Eigenmann, 1914), Knodus caquetae Fowler, 1945, K. tanaothoros (Weitzman, Menezes, Evers & Burns, 2005), K. weitzmani (Menezes, Netto-Ferreira & Ferreira, 2009) and by having 3 to 5 maxillary teeth (vs. absence in K. jacunda,one in K. caquetae, and 2 in K. moenkhausii, K. tanaothoros, K. weitzmani and K. cismontanus); from K. megalops Myers, 1929 by having 3 or 4 tricuspid teeth in the premaxillary outer row (vs. 5); from K. jacunda by having 3 to 5 maxillary teeth (vs. absence); from K. smithi (Fowler, 1913) by having 3 to 5 cusps on the teeth of the inner row of the premaxilla (vs. 7); and from K. figueredoi, K. heteresthes, K. meridae, K. mizquae, K. moenkhausii, K. victoriae, K. pasco Zarske, 2007 by having 12 circumpeduncular scales (vs. 13–14 combined). Furthermore, Knodus guajajara differs from K. victoriae by having more total vertebrae 33–35, mode 34 (vs. 30–33, mode 32).

FIGURE 4 | Knodus guajajara, CICCAA 4861, paratype, male, 31.9 mm SL Maranhão, Mearim River basin. A. Hooks on pelvic fin. B. Hooks on anal fin. (Photographed by F. P. Ottoni).

TABLE 3 | Morphometric data (N = 136) for the paratypes of Knodus guajajara from the Munim River basin. SD = Standard deviation.

Description. Morphometric data presented in Tabs. 2–3. Body comparatively small, with largest specimen examined measuring 37.7 mm SL. Greatest body depth at dorsal-fin origin. Dorsal profile of head convex from upper lip to vertical through middle portion of eye; slightly concave from this point to tip of supraoccipital spine; straight to slightly convex from posterior tip of supraoccipital spine to dorsal-fin origin; dorsal-fin base straight; slightly convex to straight from end of dorsal-fin base to adipose fin and concave from latter point to anterior dorsal-procurrent ray. Ventral profile of body convex from lower lip to anal-fin origin; straight, posterodorsally inclined along anal-fin base. Dorsal and ventral profile of caudal peduncle slightly concave.

TABLE 4 | Kimura-2-parameters pairwise genetic distances among species.