Field methods and euthanasia methods were approved by CEUA-CCS-UFRJ (Ethics Committee for Animal Use of Federal University of Rio de Janeiro; permit number: 065/18) and collections were made with permits provided by ICMBio (Instituto Chico Mendes de Conservação da Biodiversidade; permit number: 38553-7) and INEA (Instituto Estadual do Ambiente; permit number: 037/2018).

Field studies were made during five trips to the upper section of the Rio Preto drainage (May and July 2015; July 2017; September and December 2018). During this period, sporadic collections were also made in the middle and lower sections of the Rio Preto, where no specimen of T. albinotatus was found. Collections of T. albinotatus were made in 11 collecting sites (Table 1), in altitudes between 1155 and 1495 m a.s.l., comprising the Rio Preto main channel and all its main tributaries (Fig. 1). Specimen collections were made with small dip nets (40 × 30 cm) during daylight. Euthanasia followed AVMA Guidelines for the Euthanasia of Animals (Leary et al. 2020); specimens were euthanised using a buffered solution of tricaine methane sulphonate (MS-222) at a concentration of 250 mg/l, for a period of 10 min or more, until completely ceasing opercular movements. Just after collection, specimens were observed in small transparent plastic bottles to record general colour patterns and fin morphology and then euthanised, except for usually two or three individuals representing each collecting locality that were kept alive for about 5–10 hours, photographed in small aquaria and then euthanised as described above. Most specimens were fixed in 10% formalin for a period of 10 days and then transferred to 70% ethanol; specimens to be used in molecular analyses were fixed and preserved in 98% ethanol, indicated in the list of material examined (Appendix 1) as ‘DNA’ just after catalogue numbers. Specimens were later deposited in the ichthyological collection of the Institute of Biology, Universidade Federal do Rio de Janeiro, Rio de Janeiro (UFRJ). Geographical names are according to the regional popular use in Portuguese (e.g. Córrego, Ribeirão, Rio), making easier recognition of localities in the field and avoiding common equivocal generalisations when tentatively translating them to English. A complete list of specimens analysed appears in the Appendix 1.

Figure 1. 

Geographical distribution of Trichomycterus albinotatus and its colour morphs (CMs): dots, CM1; triangle, CM2; diamonds, CM3; squares, CM4. Numbers are collecting sites with DNA samples.

Total genomic DNA was extracted from muscular tissue of the right side of the caudal peduncle using the DNeasy Blood & Tissue Kit (Qiagen), according to the manufacturer’s protocol. The analyses included a fragment of the mitochondrial encoded gene cytochrome b (CYTB), which have been successfully used to delimit trichomycterine species (Costa and Katz 2021). Amplification was made through the polymerase chain reaction (PCR) method, using the primers Cytb Siluri F and Cytb Siluri R (Villa-Verde et al. 2012). Double-stranded PCR amplifications were performed in 60 μl reactions with reagents at the following concentrations: 5× GreenGoTaq Reaction Buffer (Promega), 1 mM MgCl₂, 1 mM of each primer, 75 ng of total genomic DNA, 0.2 mM of each dNTP and 1 U of standard Taq polymerase or Promega GoTaq Hot Start polymerase. The thermocycling profile was as follows: initial denaturation for 2 min at 94–95 °C; 35 cycles of denaturation for 1 min at 94°C, annealing for 1 min at 48.0–60.0 °C and extension for 1–2 min at 72 °C; and terminal extension for 4 min at 72 °C. Negative controls were used to check on contaminations. The PCR products were then purified using the Wizard SV Gel and PCR Clean-Up System (Promega). Sanger Sequencing reactions were made by Macrogen Inc. (South Korea). Sequencing chromatograms and sequences were assessed using MEGA 7 (Kumar et al. 2016). The generated DNA sequences were translated into amino acids residues to verify the absence of premature stop codons or indels using the programme MEGA 7. A list of GenBank access numbers is available in Online Resource 2.

A phylogenetic analysis was performed using a CYTB fragment, 1098 bp, for 23 specimens of T. albinotatus from 11 collecting sites (CS1–11), representing all colour morphs (CM1–4). Outgroups included single sequences of four congeners, including Trichomycterus vitalbrazili Vilardo, Katz & Costa, 2020, the sister group of T. albinotatus, the only other member of the subgenus Humboldtglanis Costa, 2021; Trichomycterus brasiliensis Lütken, 1874 and Trichomycterus mirissumba Costa, 1992, two species of the subgenus Cryptocambeva Costa, 2021, the sister group of Humboldtglanis; and Trichomycterus auroguttatus Costa, 1992, a member of the subgenus Psammocambeva Costa, 2021, the sister group of the clade comprising Cryptocambeva plus Humboldtglanis. The data were partitioned according to codon positions; the best-fitting models of molecular evolution for each partition was found using the Bayesian Information Criterion (BIC) of ModelFinder (Kalyaanamoorthy et al. 2017), implemented in IQ-TREE 1.6.11 (Nguyen et al. 2015), which found the K2P+FQ+I+G4 model for the 1^st codon position, the TN+F+I model for 2^nd codon position and the GTR+FI+G4 model for the 3^rd codon position. The dataset was analysed using Maximum Likelihood (ML) methods implemented in IQ-TREE, with two methods for assessing the reliability of internal branches: the Shimodaira-Hasegawa-like procedure support (SH-aLRT; Guindon et al. (2010)) and the ultrafast bootstrap support (UFBoot; Minh et al. (2013); Hoang et al. (2017)), using 1000 replicates and default parameters. Genetic distances were calculated to illustrate genetic diversity amongst taxa using Kimura 2-parameter (K2P) model (Kimura 1980) in MEGA 7.

Four colour morphs were found, as below described (see Fig. 1 for geographical distribution of colour morphs).

Colour morph 1 (CM1). Description: In specimens between about 30 and 60 mm SL (Fig. 2B–E), flank pale brownish-yellow, darker above longitudinal body mid-line; two horizontal rows of rounded to horizontally elongated brown to black spots, alternated with similar-shaped white spots, sometimes with faint golden iridescence. Horizontal row of pale brown spots irregularly shaped on ventral part of flank, sometimes inconspicuous. In larger adult specimens above about 65 mm SL (Fig. 2A), dark spots on flank enlarged and coalesced to form vermiculate pattern; light marks paler.

Figure 2. 

Colour morphs (CMs) of Trichomycterus albinotatus, with respective collecting sites (CSs) and collection catalogue numbers: CM1 (a–e); CM2 (f–g); CM3 (h); CM4 (i–j). a. CS3, UFRJ 11062, 66.8 mm SL; b. CS3, UFRJ 11062, 40.5 mm SL; c. CS3, UFRJ 11976, 44.3; d. CS5, UFRJ 12035, 44.9 mm SL; e. CS5, UFRJ 12035, 46.2 mm SL; f. CS6, UFRJ 12064, 45.2 mm SL; g. CS6, UFRJ 12064, 67.0 mm SL; h .CS11, UFRJ 12052, 49.2 mm SL; i. CS9, UFRJ 11670, 44.8 mm SL; j. CS10, UFRJ 12063, 71.5 mm SL.

Distribution and habitat: Areas above 1150 m a.s.l. of the upper course of the Rio Preto, the Ribeirão Santa Clara and Córrego das Cruzes drainage (CS1–6; Fig. 1). Rivers and streams about 2–15 m wide, deepest areas about 5 m deep, running in broad valleys, making the riverine environment highly exposed to sunlight. Water clear, hyaline. Common presence of large rocks favours formation of small pools where most specimens were concentrated (Fig. 3A, B). Bottom with light grey rocks and multicolour gravel, predominating white, light grey and light brownishyellow colours.

Figure 3. 

Habitats of Trichomycterus albinotatus: a. Upper Rio Preto, the type locality of T. albinotatus; b. Upper Córrego das Cruzes; c. Cachoeira do Marimbondo, upper Rio do Marimbondo subdrainage; d. Cachoeira da Muralha, upper Córrego do Alcantilado subdrainage; e. Cachoeira das Antas, upper Ribeirão das Flores subdrainage.

Colour morph 2 (CM2). Description: Similar to CM1, except for white marks irregularly shaped and dark marks highly coalesced even in smaller specimens (Fig. 2F). In a single specimen found above 65 mm SL, white spots absent (Fig. 2G).

Distribution and habitat: Upper Rio do Marimbondo, altitude about 1435 m a.s.l. (CS6; Fig. 1). River about 5–10 m wide, deepest areas about 3 m deep, running in broad valley, riverine environment highly exposed to sunlight. Water clear, greenish hyaline. Common presence of large rocks favours formation of small pools below waterfalls where most specimens were concentrated (Fig. 3C). Bottom predominantly with light grey rocks of different sizes.

Colour morph 3 (CM3). Description: Similar to CM1, but white spots on body mid-line highly coalesced, usually with intense golden iridescence (Fig. 2H).

Distribution and habitat: Rio das Flores subdrainage at about 1220 m a.s.l. (Fig. 1; CS11). Rivers about 5–10 m wide, deepest areas about 2 m deep, running in broad valley, riverine environment highly exposed to sunlight. Water clear, yellowish hyaline. Rocks small, rare small waterfalls where most specimens are concentrated (Fig. 3E). Bottom predominantly with small light grey rocks, light grey to light yellow gravel and sand.

Colour morph 4 (CM4). Description: Flank and head pale yellow, with dark brown to black spots (Fig. 2I–J); flank dark spots highly coalesced in larger specimens (Fig. 2J), forming interconnected vermiculate pattern, flank white spots almost inconspicuous in smaller specimens (Fig. 2I), always absent in specimens above about 65 mm SL (Fig. 2J).

Distribution and habitat: Upper section of the Córrego do Alcantilado (CS7–10; Fig. 1), at altitudes between 1155 and 1320 m a.s.l. Stream about 2–5 m wide, deepest areas barely reaching about 2 m deep, running in narrow and steep valley, with numerous small waterfalls (Fig. 3D); riverine environment poorly exposed to sunlight. Water clear, yellowish hyaline. Bottom predominantly composed of dark grey rocks.

The phylogenetic analysis highly supported T. albinotatus as a unique lineage, with maximum values for both SH-aLRT and UFBoot (Fig. 4). The tree topology did not support any correlation between colour morphs and lineages, with colour morphs not forming unique lineages. Genetic distances amongst haplotypes were low (0.1–0.4%). The eleven haplotypes found amongst the 23 sequenced individuals were scattered in different localities. H1 was the most common haplotype, present in seven analysed specimens collected in seven separate localities and exhibited all the four colour morphs.

Figure 4. 

Phylogenetic relationships amongst 23 specimens of Trichomycterus albinotatus and three outgroups, inferred by Maximum Likelihood analysis performed in IQ-TREE using a cyt b fragment 1098 bp. Numbers close to nodes are support values: SH-aLRT support (%) and ultrafast bootstrap support (%). CS1–11, collecting sites 1–12; CM1–4, colour morphs 1–4; H1–11, haplotypes 1–11. The most external outgroup is not represented in the figure.

High levels of intraspecific colouration polymorphism have been recorded in taxonomical papers on some trichomycterines, including both discrete and indiscrete colouration variation in syntopic specimens (e.g. Arratia et al. (1978); Sarmento-Soares et al. (2005); Lima et al. (2008)). On the other hand, according to Arratia and Menu-Marque (1981) and Arratia (1983a), some Andean trichomycterine species exhibit distinct colour morphs in disjunct habitats with different river substrates. In the present study, we found specimens of T. albinotatus exhibiting different colour morphs in disjunct areas of the upper Rio Preto drainage, with slightly different ecological conditions (Fig. 2; Table 2). In the most extreme case (compare Fig. 2C with Fig. 2I for specimens about 45 mm SL and Fig. 2A with Fig. 2J for specimens about 65–70 mm SL), specimens found in habitats with light substrate, situated in wide valleys highly exposed to sunlight, had a colour pattern consisting of alternating white and brown to black spots (CM1, Fig. 2A–C). Contrastingly, specimens found in a stream with dark grey rocks on the bottom, situated in a narrow steep valley weakly exposed to sunlight were darker, with dark spots elongated and coalesced and white spots inconspicuous or absent (CM4, Fig. 2I–J). It is remarkable that, during daylight field studies, CM1 specimens when swimming close to multicolour gravel substrate were often almost invisible, showing that CM1 may favour camouflage in this kind of habitat. These data suggest that the existence of different colour morphs may be related to specific kinds of habitat, a hypothesis to be tested only after detailed inventory on physical and chemical factors of the riverine habitats being available.

In all colour morphs of T. albinotatus, white marks on the flank were less conspicuous or had completely disappeared, whereas dark marks have expanded their extent in individuals above 65 mm SL (Fig. 2A, G, J). This ontogenetic colour change may be related to field evidence indicating that specimens about 60 mm or less are typically diurnal, whereas larger ones are apparently nocturnal. The presence of rows of alternating dark and light spots on the body side is a common colour pattern occurring in diurnal trichomycterids living in sandy or gravel substrate, which contribute to making them little visible for predators (Zuanon and Sazima 2004). For example, in T. auroguttatus, a typical diurnal psammophilic species sympatric to T. albinotatus (Costa 2021), there are alternating dark brown and golden spots on the flank (Costa 1992). On the other hand, different species having cryptic habits during daylight have dark colouration and no bright mark (Costa et al. 2020b). A similar ontogenetic change was recorded for T. vitalbrazili, the sister group of T. albinotatus (Vilardo et al. 2020).

Little is still known about natural history of trichomycterines; therefore, it is not possible to infer securely if the synchronic change of colouration and period of activity in T. albinotatus and T. vitalbrazili corresponds to a unique evolutionary event or if it is a widespread condition amongst trichomycterines. Miranda-Ribeiro (1905) first reported conspicuous differences in the colouration of juvenile and adult specimens of trichomycterines. Since then, several papers have provided evidence of ontogenetic colouration change in Trichomycterus and the closely-related genus Cambeva (Costa 1992; Barbosa and Costa 2008, 2010b; da Silva et al. 2010; Ferrer and Malabarba 2013; Reis et al. 2020; Costa et al. 2021, 2022b). Amongst these studies, only da Silva et al. (2010) have demonstrated that small and large individuals of Cambeva iheringi (Eigenmann, 1917) besides having a distinct colour pattern, are also found in different habitats. According to these authors, spotted juveniles of C. iheringi live partially or totally buried in light sandy substrate, whereas dark-coloured larger adults are found in areas with dark stones. Arratia (1983b) reported juveniles and small adults of three Chilean trichomycterines living in shallower and more exposed habitats than larger adults, but no mention was made of possible ontogenetic change in colouration. Furthermore, recent studies in species of the trichomycterid subfamilies Copionodontinae and Trichogeninae indicate that ontogenetic changes in the activity period is followed by colouration change (Sazima 2004; Zanata and Primitivo 2013). Sazima (2004) noted that juveniles and small adults of Trichogenes longipinnis Britski & Ortega, 1983 are more active at the daytime, whereas large adults with darker colours are mainly seen at night. Similarly, Zanata and Primitivo (2013) reported diurnal nektonic habits in small juveniles of Copionodon pecten de Pinna, 1992, as well as nocturnal benthonic habits in sub-adults and adults, which may exhibit a darker coloured phenotype. Therefore, since copionodontines and trichogenines together form a clade distantly related to trichomycterines (Katz et al. 2018), it is possible that synchronic change of colouration and habits during ontogeny have occurred independently in the clade comprising copionodontines and trichogenines and in the clade comprising T. albinotatus and T. vitalbrazili.

This study firstly reported different colour morphs of T. albinotatus inhabiting geographically disjunct areas, which could induce taxonomists to recognise them as different species. Although the occurrence of geographically disjunct colour morphs is possibly uncommon amongst trichomycterids, data on colouration polymorphism provided by Arratia & Menu-Marque (1981) and Arratia (1983a) suggest that colouration may be vary according to local environmental conditions in different trichomycterid lineages, attesting against the use of colouration characters in trichomycterid systematics.

Similarly, the use of colouration alone to distinguish sympatric species may induce misidentifications. For example, sympatric juveniles of T. itatiayae Miranda Ribeiro, 1906 and T. nigroauratus Barbosa & Costa, 2008, two species belonging to different subgenera (Costa 2021), have similar colouration, making it difficult to distinguish them in the field (Barbosa and Costa 2008), as well as the non-closely related, sympatric T. immaculatus (Eigenmann & Eigenmann, 1898) and T. quintus Costa, 2020 that share a similar overall black colouration, are only distinguishable under careful examination (Costa et al. 2020b).

These data do not diminish the value of colour pattern characters to diagnose species and trichomycterine clades. For example, even with the occurrence of chromatic polymorphism herein reported for T. albinotatus, the presence of white marks on the flank of juveniles uniquely shared by T. albinotatus and T. vitalbrazili amongst trichomycterines, clearly corroborated monophyly of this clade formally named as subgenus Humboldtglanis Costa, 2021, which is highly supported in a phylogenetic study integrating morphology and DNA sequences (Costa 2021). Similarly, apomorphic colour patterns are shared by trichomycterine clades corroborated by both osteological and molecular data (Costa 2021; Costa and Katz 2021), thus demonstrating that colour patterns may be effectively used for diagnosing trichomycterine species and groups. These data in fact show that colouration should not be discarded a priori as evidence of species limits or phylogenetic relationships in trichomycterines, as well as colouration alone should not be used to infer species limits, being necessary to reinforce species delimitation hypotheses using additional morphological evidence (e.g. osteological characters) or molecular data.