First DNA barcode library for the ichthyofauna of the Jos Plateau (Nigeria) with comments on potential undescribed fish species

Fieldwork, study sites and species identification

Sampling was carried out on several field trips to the Jos Plateau during the dry season (December 2013 and January-April 2014). Fishes were collected at 37 strategically selected sampling locations in 17 major rivers on the Jos Plateau (Table S1 provides detailed locality information, Figs. 1 and 2). Fishes were captured using different methods (cast net, gill net, frame net, fish traps) depending on the size and habitat of the site. Freshly caught specimens were euthanized with an overdose of anaesthetic (Benzocaine, MS-222). Specimens representative of all taxa from each collection at each sampling site were individually tagged and pectoral fin clip sampled. The whole specimens were photographed after-which they were fixed in 5% formalin and eventually preserved in 70% ethanol after being treated in changes of water followed by different concentrations (20%, 40% and 60%) of ethanol according to Neumann (2010). Tissues were placed immediately into individually labelled tubes filled with 96% ethanol. Specimens were exported on the 17^th of September, 2014 and subsequently deposited in the fish collection of SNSB-ZSM (Germany). We followed all applicable international and national guidelines of animal use and ethical standards for the collection of samples. The permit to survey the fishes of the Jos area was obtained from the GWOM SOPP palace, Riyom, Plateau State. After a preliminary identification of specimens in the field to a genus level, preserved specimens were subsequently assigned to a species using standard identification keys (Paugy, Leveque & Teugels, 2003; Stiassny, Teugel & Hopkins, 2007). When required, additional literature (i.e. original descriptions) were consulted allowing species identification and the clear identification of diagnostic characteristics. This was especially necessary for the identification of the candidate species (following the terminology regarding potentially undescribed species of (Padial et al., 2010; Vieites et al., 2009)) belonging to the genera Enteromius and Clarias and their comparison to potentially similar species. When morphometric measurements were required, we measured those following protocols in the identification keys using a manual calliper. Important meristic traits were examined using a stereomicroscope (Leica MZ6) as well as with a digital X-ray using a Faxitron Ultrafocus LLC X-ray unit. However, some of the collected specimens could not be assigned to any described species known from West Africa (please see further below for a comprehensive overview of the newly discovered candidate species).

Notes on collection sites (see Fig. 2)

Most of the collection sites were situated in perennial or annual headwaters of major rivers in Nigeria. Generally, they originate as small streams before they flow over long stretches of fine-grained alluvial soils with few rocks. After a substantial increase in stream size, they reach the margin of the high plateau before descending, often in rapids and cascades, to the lowland plains below. On the plateau, water flow is usually moderate, whereas in rapids and waterfalls it may be torrential. Vegetation along the river banks provides partial shading and is best characterized as a savanna-forest mosaic. None of the sampled locations was pristine; all were impacted by farming, domestic waste deposition, industrial discharges or by mining.

Water chemistry (Table S3)

The physico-chemical characteristics of the Assop, N’gell, Rukuba, and Tahoss rivers were assessed following standard procedures. The water temperature was measured in situ using mercury-in-glass bulb thermometer with calibration range (−10 °C to 110 °C). Samples for the analysis of physico-chemical parameters were collected in clean 2 litre plastic bottles. However, dark reagent bottles were used for Biochemical Oxygen Demand (BOD₅) sample collections and kept in a dark cupboard for 5 days for subsequent analysis. The pH, conductivity, and total dissolved solids (TDS) of each sample were determined using a PCE-PHD1 multi-parameter meter. The analytical determinations of the physico-chemical water quality parameters were carried out within the holding time of each parameter (Ademoroti, 1996; APHA, 1995; Golterman, Clymo & Ohnstadt, 1978). The BOD₅ was determined by iodiometric titration (Golterman, Clymo & Ohnstadt, 1978). Colour was investigated using a colourimeter (Jenway 6051 model) standardized with a set of potassium chloroplatinate-cobalt solutions as Pt-Co standards, while turbidity was assessed using a turbidimeter (Ademoroti, 1996). Sulphate (SO₄²⁻) and nitrate (NO₃⁻) were determined by spectrophotometric methods (Ademoroti, 1996). Magnesium (Mg²⁺) was determined by the complexio-metric titration method using Na₂EDTA (Golterman, Clymo & Ohnstadt, 1978). All recommended quality control (QC) and quality assurance (QA) measures were taken for each determination.

Laboratory protocols

For DNA extraction, we subsampled the ethanol-preserved fin clips and transferred them to 96-well plates. These plates were sent to the Canadian Center for DNA Barcoding (CCDB, Guelph, Canada) for DNA extraction, PCR amplification, and Sanger sequencing (all protocols are available online: ccdb.ca/resources/). PCR amplification of all samples employed the fish primer pair C_FishF1t1/C_FishR1t1 (Ivanova et al., 2007) for the CO1 (mitochondrial cytochrome c oxidase subunit 1) barcode fragment. The same primers were used for subsequent bidirectional Sanger sequencing reactions. The resulting sequence data and trace files were uploaded to barcode of life database (BOLD) (Ratnasingham & Hebert, 2007) as well as all corresponding metadata voucher information (i.e., locality data, altitude, taxonomic classification, habitat images, primer information etc.) and are publicly accessible on BOLD in dataset DS-NGAJO (DOI 10.5883/DS-NGAJO).

Molecular data anaylsis

We used the “Sequence Analysis” toolbox on BOLD and restricted analysis to sequences >500 bp. MUSCLE (Edgar, 2004) was employed to align the sequences while the sequence divergences (i.e., mean and maximum intraspecific variation as well as the minimum genetic distance to the nearest-neighbour species) were calculated using the ‘Barcode Gap Analysis’ tool on BOLD. This analysis employs the Kimura-2-Parameter (K2P) distance metric (Puillandre et al., 2012). As well a neighbour-joining (NJ) tree was constructed to graphically represent the BIN clusters. Within BOLD, closely similar COI barcode sequences are assigned a globally unique identifier, termed a “Barcode Index Number” or BIN (Ratnasingham & Hebert, 2013). The ‘‘BIN Discordance’’ report on BOLD was used to reveal cases where BINs were shared between species clusters or restricted to certain species clusters (BIN Discordance and BIN Sharing). In addition, we obtained for each BIN recovered on the Jos Plateau the nearest BIN available on BOLD (which includes data deposited on GenBank) along with corresponding divergence information directly from the individual BIN records (see Table S4). This was done by screening the BOLD database for all available CO1 sequences for each genus (with a minimum length of 500 bp) using the “Record Search” tool, including public records mined from GenBank. In addition, we created additional and more comprehensive CO1 DNA barcode sequence datasets for all genera for which putatively undescribed species (see below) appeared to be present on the Jos Plateau region. This was equally done by screening the BOLD database for all available CO1 sequences for each genus (with a minimum length of 500 bp) using the “Record Search” tool. Subsequently, we constructed a NJ tree for each genus as indicated above.

DNA barcode library

In total, 176 CO1 DNA barcode sequences were generated with lengths >635 bp (1 shorter barcode-444 bp, was excluded from further sequence analysis). These sequences represented 20 of the 23 fish species collected on the Jos Plateau. Three species lacked barcode coverage because of suboptimal tissue preservation and low sample size: Chiloglanis cf. benuensis Daget & Stauch, 1963, Clarias cf. gariepinus Burchell, 1822 and Marcusenius mento Boulenger, 1890. Only a single barcode was obtained for five species (Heterobranchus longifillis, Labeo sp. Assop, Labeobarbus bynni, Synodontis violaceus, and Poecilia reticulata), whereas the other species all had two or more sequences. Overall, CO1 amplification was successful for 92% of the 190 submitted specimens. The sequence divergence values for the CO1 barcode region (Table S4) show that most species of the Jos Plateau possessed very low intraspecific genetic distance values.

An NJ tree (Fig. 3) showed that barcodes reliably discriminated most species as all conspecific specimens were recovered as monophyletic lineages assigned to a single BIN (19 BINs in total). Oreochromis niloticus and Sarotherodon galilaeus shared a BIN, but formed two different haplotype clusters with a low sequence divergence (see Fig. 3). These results are similar to those in an earlier barcoding study on the freshwater fishes of south-eastern Nigeria (Nwani et al., 2011) which found very low genetic divergence between O. niloticus and Sarotherodon galilaeus boulengeri. While another barcoding study on the ichthyological fauna of the Nile recovered a low genetic divergence between Oreochromis aureus (Steindachner, 1864) and S. galilaeus (Ali, Ismail & Aly, 2020). The clustering of S. galilaeus within Oreochromis might either point to regular mis-identification of this species, which however appears rather unlikely, or a taxonomic misassignment of S. galilaeus in the genus Oreochromis. However, recent nuclear based studies support the monophyly of the genus Oreochromis while Sarotherodon clearly appears to be polyphyletic (Dunz & Schliewen, 2013; Ford et al., 2019). Overall, the afore-mentioned studies in combination with our results alternatively suggest a cytonuclear discordance in respect of the phylogenetic position of S. galilaeus, which might be the result of past hybridization events. Indeed, hybridization between Oreochromis and Sarotherodon has been previously suggested to have taken place (see Ford et al., 2019; Nagl et al., 2001). Further, it is known that both genera hybridize in captivity (Otubusin, 1988) and are commonly farmed in aquaculture facilities throughout Nigeria (Ayinla, 2007). In any case, the biological reason for the BIN sharing between the O. niloticus and S. galilaeus on the Jos Plateau and elsewhere remain uncertain and would need to be critically tested with appropriate genomic datasets.

Notes on new candidate species from the Jos Plateau

The morphological investigation of the collected specimens as well as the subsequent COI analyses suggest the presence of at least five candidate species, four Cyprinidae, and one Clariidae. Among the potentially new cyprinid species, two belong to the genus Enteromius (Cope, 1867). This genus was recently revalidated by (Yang et al., 2015) for the small diploid African barbs formerly placed in the genus Barbus (Daudin, 1805), but is known to be polyphyletic (Yang et al., 2015). Although a comprehensive taxonomic re-evaluation of the diploid African barbs is needed, this interim generic assignment is currently generally accepted (Englmaier, Tesfaye & Bogutskaya, 2020; Hayes & Armbruster, 2017; Martin & Chakona, 2019; Skelton, 2016).

The first species, herein referred to as Enteromius sp. Silver (see Fig. S2), was widespread on the Jos Plateau, being present at most collection sites. It is phenotypically similar to Enteromius callipterus (Boulenger, 1907), a widely distributed West African barb originally described from Cameroon but differs by its possession of a faint greyish longitudinal band (vs no lateral colour pattern for E. callipterus). In addition, the dorsal spot in the dorsal fin is less pronounced, i.e., irregular and not deep black but rather greyish with translucent orange (background colour of the dorsal fin), as compared with E. callipterus (Leveque, 2003). In a NJ tree based on all available CO1 sequences of Enteromius on BOLD (see Fig. S5) Enteromius sp. Silver represents a distinct mitochondrial lineage not clustering with E. callipterus. Instead, it was sister group to a cluster including different Enteromius species occurring in the Congo drainage i.e., Enteromius brazzai (Pellegrin, 1901) and the Zambezi drainage i.e., Enteromius radiatus (Peters, 1854).

The second candidate species, referred herein as Enteromius sp. Gold (see Fig. S2) was collected from four rivers (N’gell, Delimi, Shen Fusa, Rukuba). It is a comparatively small species easily recognized by its yellowish to golden body coloration and small irregular blackish spots forming a more or less confluent longitudinal band on the sides which terminates before the base of the caudal fin in a slightly larger blackish spot. Morphologically, Enteromius sp. Gold could not be assigned to any Enteromius species known from West Africa or Lower Guinea based on standard keys (Leveque, 2003; de Weirdt et al., 2007). Interestingly, a NJ tree based on all CO1 sequences for Enteromius on BOLD (see Fig. S5) revealed that Enteromius sp. Gold was a sister taxon to Enteromius cf. macrotaenia (Worthington, 1933), a species known from south-eastern Africa. The sister group of this cluster included Enteromius lukusiensis (David & Poll, 1937), Enteromius greenwoodi (Poll, 1967), and the goldie barb Enteromius pallidus (Smith, 1841), which is endemic to the eastern Cape Fold Ecoregion of South Africa (Martin & Chakona, 2019). These results suggest the closest relatives of Enteromius sp. Gold are found in the southern parts of Africa, raising interesting questions on its biogeographic history (i.e., when and how it colonized the Jos Plateau area).

The third potentially new cyprinid species, referred herein as Labeo sp. Assop, was only collected from the Assop River (see Fig. S2). It shows affinities with Labeo parvus Boulenger, 1902 which is widespread on the Jos Plateau. A NJ tree including all available CO1 sequences for Labeo on BOLD (see Fig. S6) revealed that Labeo sp. Assop was a sister taxon to a clade including Labeo lukululae Boulenger, 1902 Labeo quadribarbis Poll & Gosse, 1963, and Labeo sp. aff. rectipinnis Tshibwabwa, 1997. Interestingly, and similar to the case of Enteromius sp. Gold, all three species forming the sister group to Labeo sp. Assop are neither present in the Niger/Benue drainage system nor in the Chad system but are restricted to the Congo drainage system.

The fourth potentially new cyprinid species, referred herein as Labeobarbus sp. Assop, is the second candidate species collected only from the Assop River (see Fig. S3A). It could not be assigned to any known Labeobarbus species found in western Africa based on morphological data and following the key in Leveque (2003). The collected specimens were recovered as sister to Labeobarbus brevispinis (Holly, 1927), a species known from Lower Guinea but as well from the Benue River drainage (de Weirdt et al., 2007), in a NJ tree based on all available CO1 sequences on BOLD (see Fig. S6).

The last candidate species, referred herein to as Clarias sp. White dots (see Fig. S4), was found in four rivers (N’gell, Gurum, Kassa, Rukuba). It could not unambiguously be assigned to any described Clarias species known from Western Africa based on standard morphological keys (Teugels, 2003; Teugels et al., 2007). Nevertheless, it is phenotypically similar to Clarias agboyienesis Sydenham, 1980, a species known from West African coastal basins including the lower course of the Niger River. The latter species is, however, characterized by a very light coloration with yellowish brown flanks and a light grey belly whereas the body ground coloration of Clarias sp. White dots is dark with larger specimens having small whitish to yellowish spots on the flanks and on unpaired fins. Interestingly, the sequenced specimens of Clarias sp. White dots clustered with sequences of Clarias gabonensis Günther, 1867 in the NJ tree based on all CO1 sequences for Clarias on BOLD (see Fig. S7). The coloration of C. gabonensis is either uniform brownish yellow or marbled with small yellowish spots (Hanssens, 2007; Teugels et al., 2007) which shows some correspondence to the coloration pattern of Clarias sp. White dots, but it has a distinctively shorter head. Furthermore, C. gabonensis occurs in the Congo basin and different drainage systems of Lower Guinea, e.g., the Ogowe, Noya, Kouilou and Chiloango rivers (Hanssens, 2007; Teugels et al., 2007) and it was only recently recorded from Nigeria (Nwani et al., 2011). However, the taxonomic assignment of the Nigerian specimens of Clarias gabonensis might be incorrect and it should be carefully investigated, since the identification of Clarias species is often challenging. Indeed, the available sequences of C. gabonensis were dispersed in distinct clusters in our NJ tree (see Fig. S7), indicating probable misidentifications or the presence of cryptic species. Our sequences of Clarias sp. White dots group with some Nigerian C. gabonensis in one of the larger clades identified as C. gabonensis. Taken together it would be premature to assign our specimens of Clarias sp. White dots to Clarias gabonensis until more thorough morphological and genetic investigations are made within this species complex.

A thorough taxonomic investigation and the formal description of these candidate species is in progress and will be presented in a follow up study.

In addition to these five putatively new species, we observed substantial phenotypic variation in two species. First, in the cichlid, Coptodon zillii, which is widespread on the Jos Plateau and occurs in two color morphs, morph 1 and morph 2. Both morphs occurred sympatrically at one single location (Delimi River). Although differences between these morphs is modest, the flank coloration of one morph is predominantly yellowish (Morph 1) whereas in the other morph is whitish (Morph 1) (see Fig. S1). Furthermore, morph 1 has a pale green upper lip and and a whitish lower lip while morph 2 has greenish to brownish upper lips and blueish lower lips. Members of the two morphs formed two distinct mitochondrial lineages with low genetic divergence (see Fig. 3).

Specimens of Enteromius perince (Rüppell, 1835) also showed clear intraspecific variation. Most specimens showed the colouration typical of this species (i.e., silver body ground coloration with three spots aligned along the mid-lateral line), but a few specimens had additional spots. These individuals resemble the so-called “lepidus form” (Leveque, 2003), but no CO1 barcodes were recovered from them so their divergence from the nominate form is uncertain.