Research Article |
Corresponding author: Mariam I. Adeoba ( mariamsalami@yahoo.co.uk ) Academic editor: Nina Bogutskaya
© 2018 Mariam I. Adeoba, Ronny Kabongo, Herman Van der Bank, Kowiyou Yessoufou.
This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Citation:
Adeoba MI, Kabongo R, Van der Bank H, Yessoufou K (2018) Re-evaluation of the discriminatory power of DNA barcoding on some specimens of African Cyprinidae (subfamilies Cyprininae and Danioninae). ZooKeys 746: 105-121. https://doi.org/10.3897/zookeys.746.13502
|
Specimen identification in the absence of diagnostic morphological characters (e.g., larvae) can be problematic even for experts. The goal of the present study was to assess the performance of COI in discriminating specimens of the fish family Cyprinidae in Africa, and to explore whether COI-phylogeny can be reliably used for phylogenetic comparative analysis. The main objective was to analyse a matrix of COI sequences for 315 specimens from 15 genera of African Cyprinidae using various distance-based identification methods alongside multiple tests of DNA barcode efficacy (barcode gap, species monophyly on NJ tree). Some morphological and biological characters were also mapped on a COI-phylogeny reconstructed using Maximum Parsimony. First, the results indicated the existence of barcode gaps, a discriminatory power of COI ranging from 79 % to 92 %, and that most nodes form well-supported monophyletic clades on an NJ tree. Second, it was found that some morphological and biological characters are clustered on the COI-phylogeny, and this indicates the reliability of these characters for taxonomic discrimination within the family. Put together, our results provide not only an additional support for the COI as a good barcode marker for the African Cyprinidae but it also indicate the utility of COI-based phylogenies for a wide spectrum of ecological questions related to African Cyprinidae.
BRONX algorithm, character variations, COI, specimen identification
Cyprinidae is the most diverse family of freshwater fishes (Nelson et al. 2006;
Traditionally, external morphological and osteological characteristics have been used to differentiate species within the subfamilies Cyprininae and Danioninae (
The taxonomy of the family has been a topic debated in several studies (e.g.,
Similarly, the recent molecular and morphological work of
The present study uses a broader sampling of the African Cyprinidae and integrates morphology and ploidy data to further assess the effectiveness of DNA barcoding in discriminating specimens within the family. Specifically, the aim was to: (i) test the reliability of COI as a DNA barcode for the African Cyprinidae based on barcode gap, various distance methods, and the Rosenberg test of species monophyly; and (ii) map six traits including five morphological characters and ploidy level onto a COI-based phylogeny of the African Cyprinidae.
First, 584 COI sequences of the African Cyprinidae specimens were retrieved from the Barcode of Life Database (BOLD; www.boldsystems.org) and GenBank/EBI (www.ncbi.nlm.nih.gov/nuccore). Some of the sequences from BOLD had been generated from our group (African Centre for DNA Barcoding) (Suppl. material
The 55 new COI sequences mentioned above were generated following the manufacturers’ recommended protocol developed from NucleoSpin® Tissue kit (Macherey- Nagel). The sequence amplification (PCR) was done in accord with
After the amplification, PCR products were visualised on 1.5% agarose gels. Visible products were cleaned using silica column kits, viewed again on agarose gels, and selected for cycle sequencing. Sequencing of COI region was done following the standard protocols of the Canadian Centre for DNA Barcoding (CCDB). Sequences were aligned using Multiple Sequences Comparison by Log-Expectation (MUSCLE vs. 3.8.31;
Because some DNA sequences available on public repositories are not reliable (
All barcoding analysis was conducted in the R package SPIDER (species identity and evolution in R) vs. 1.1-1 (
The discriminatory power of the COI gene was tested with three methods: Best Close Match, Near Neighbour and the BOLD identification (threshID) (
To test for species monophyly, a tree based analysis using
A DNA matrix of 315 COI aligned sequences and three outgroups (Suppl. material
Information related to morphological characters and ploidy levels were collected from several sources and presented in Suppl. material
The length of the aligned COI matrix was 652 bp with the following base composition: A: 25.9 %, C: 26.8 %, G: 18.2 % and T: 29.1 %. The interspecific genetic distances (K2P) ranged from 0 to 0.30 (median = 0.15) and are larger than the intraspecific genetic distances (range: 0 – 0.02; median = 0.001; p < 0.001; Figure
Evaluation of barcode gap in the dataset. Boxplot of the interspecific (inter) and intraspecific genetic (intra) distances, indicating the existence of a barcode gap, i.e., interspecific distance is larger than intraspecific distance. The median is indicated by the horizontal line and the range as the vertical dashed lines and outliers by bold vertical lines.
Evaluation of barcode gap in the dataset. Line plot of the barcode gap for the 315 Cyprinidae individuals. The black lines indicate where the smallest interspecific distance is longer than the longest intraspecific distance (bottom of line value), thus showing the existence of a barcode gap. The red lines show where this pattern is reversed.
Determination of the threshold genetic distance for species identification. False positive (grey) and false negative (black) identification error rates summed across a range of distance thresholds from 0.01 to 1.9 %. The cumulative error plot indicates the transition between intraspecific and interspecific distances, the genetic distance corresponding to the least cumulative error (1.51 %) showing the appropriate threshold value for the dataset.
In addition, the result presented in Figure
Tests of barcoding identification accuracy with numbers (N) and percentages (%) of each score Near neighbour, Best Close Match and Bold criteria.
Methods | Near neighbour | Best Close match | Bold criteria | |||||||
---|---|---|---|---|---|---|---|---|---|---|
Scores | False | True | Ambiguous | Correct | Incorrect | No ID | Ambiguous | Correct | Incorrect | No ID |
N | 25 | 290 | 10 | 278 | 17 | 10 | 57 | 250 | 1 | 7 |
(%) | 7.9 | 92.1 | 3.2 | 88.2 | 5.4 | 3.2 | 18.1 | 79.4 | 0.3 | 2.2 |
Although COI is a universally accepted DNA barcode for animal groups (
The results presented in this work confirm that COI can be reliably used from a barcode perspective to distinguish between specimens of the African Cyprinidae in a dataset of 315 specimens representing 14 out of the 24 (58 %) recognised genera in Africa. For example, a significant barcode gap was found irrespective of the methods used, and this has also been reported for Cyprinidae of other geographic regions (e.g.,
For example, the high level of COI discrimination is further supported by the test of species monophyly, a test that resulted in strongly supported clades based on
This evidence of monophyly accords with the morphology-based taxa delimitation as we found that some morphological characters and ploidy levels clustered within some clades along the phylogeny. Such characters that clustered within clades include, for example, the number of anal soft rays and presence/absence of barbels for the tribe Smiliogastrini, the fish length (21–40 cm) for the tribe Labeonini and the tetraploidy for Barbinini. Such clustering on the COI-phylogeny is evidence not only for COI as DNA barcoding of some African Cyprinidae 11 a good barcode for the family Cyprinidae but also that COI-phylogeny can be used for a comparative phylogenetic analysis. Only the tribe Labeonini sensu Rainboth 1991 (
Overall, the existence of DNA barcode gap and a high discriminatory power, as well as the high level of monophyly give support to the use of COI as a reliable DNA barcode for African Cyprininae and Danioninae. Several studies have examined the phylogeny of this family at subfamily and genus levels using both mitochondrial and nuclear genes (
Kowiyou Yessoufou acknowledges the South Africa’s National Research Foundation (NRF) for funding (Grant No: 103944) as well as La Société Botanique de France. He also acknowledges a start-up grant from the University of Johannesburg (URC Grant: 073450). We also thank the editor and anonymous reviewers for their comments.
Table S1
Data type: occurence
Explanation note: Taxonomy of African Cyprinidae and geographic origins.
Table S2
Data type: occurence
Explanation note: List of 315 Cyprinidae specimens analysed in this study. Full names, voucher information, and geographic origins are presented. Accession numbers for ACDB (African Centre for DNA Barcoding at the University of Johannesburg), GenBank, and BOLD are also included.
Table S3
Data type: measurement
Explanation note: Matrix of the morphological characters used for character mapping along a phylogeny. Details of the characters (1–6) and sources are provided at the bottom of the table..