Fig 1.
Distance tree of snakehead barcodes.
Neighbour joining tree based on HKY distances of 777 channid coxI sequences. Individuals have been collapsed into 61 BINs and 29 potential new BINs (indicated as clades 1–29 and highlighted in dark grey); see Table 1. The Parachanna africana clade is highlighted in light grey. Numbers in brackets behind species names refer to number of individuals from this study / number of individuals from GenBank, followed by BIN or clade designation. BINs from BOLD not previously reported by [32] are indicated in brackets with "new" (BOLD accessed March 31, 2016). Red dots indicate clades containing misidentified or incomplete identified specimens; see Table 2 and the following comments: (a) Protopterus annectens BIN containing five lungfish specimens, (b) three Pa. africana and two Pa. obscura included, (c) one Pa. sp included, (d) two Pa. insignis included, (e) nine Channa sp. included, (f) one C. cf. marulius included, (g) ten C. marulius with accession numbers EU342199-EU342200, HM117192-HM117196, KJ937341, KJ937348, KJ937388 included. These might be C. pseuomarulius, but were not counted as misidentifications since this species was only recently revalidated by Britz et al. [48], (h) includes an exceptionally long branch for a single specimen (accession number JX983243 shown as a grey bar) probably due to sequencing errors, (i) one C. cf. nox included, (j) five C. marulius included, (k) one C. orientalis included, (l) BIN number is for unpublished C. orientalis in BOLD, (m) five C. orientalis included, (n) six C. orientalis included, (o) six C. orientalis included, (p) 17 C. orientalis included, (q) misidentification should be C. bleheri see Table 2, (r) one C. stewartii included, (s) one C. cf. stewartii included, (t) five C. barca included, (u) misidentification should be C. andrao see Table 2, (v) three C. gachua included, (w) BOLD ID showed 97.39% similarity with ACS3540, (x) five C. stewartii included, (y) BOLD ID showed 97.98% similarity with AAC3925, (z) one C. gachua included. For clades composed of misidentified specimens only, the original species designation was kept and these clades are underlined (ACB7513 and ACB8348).
Fig 2.
BEAST analysis using a coalescence prior. Species delimitations based on BIN, GMYC single, GMYC multiple, and PTP thresholds are indicated by black bars. The same figure with species names is given in S4 Fig.
Table 1.
Summary of channid species and their BIN and clade assignment following the clade order in Fig 1.
Species names according to [32] if different from this study are given, BINs reported by [32] are indicated and those species with multiple BINs in the [32] study and in this study are indicated.
Table 2.
Summary of channid misidentifications and incompletely identified channid specimens with their BIN assignment, GenBank accession numbers and number of specimens found in GenBank, with how many specimens have been used and / or generated by [32].
Fig 3.
Summary of species delimitation.
Cumulative number of channid species from the year 1758 to 2016 and results of channid species numbers estimated by different species delimitation methods. For the PTP analyses the species number is given (dot) and the mean and minimum and maximum range is given based on the analyses of 100 ML bootstrap trees. GMYCs and GMYCm the number of species and the confidence interval are given. For the bGMYC analyses, the mean and minimum and maximum range is given. The horizontal lines indicate different species counts: a) 38 valid species, b) plus seven undescribed species included in this study (see Fig 1): Channa sp. Assam, C. sp. Bhutan foothills, C. sp. Rakhine Yoma, Channa sp. Northeast India, Channa sp. Tenasserim, Channa sp. Mogaung and P. sp DRCongo c) plus an additional eight potential species based on a conserved estimate of additional intraspecific diversity within C. bankanensis (two clades in total), C. marulius (three clades in total), C. striata (four clades in total), C. gachua (at least three clades), totaling a conservative estimate of 53 channid species.
Fig 4.
Dotplot illustrating the presence/absence of local barcode gaps. For each individual the maximum intraspecific distance is plotted against the minimum interspecific distance. The "species" groupings were A) current taxonomy with the exception of species with prominent intraspecific clades that were split into distinct units (Channa bankanensis, C. gachua, C. marulius, C. striata), B) according to the 90 BINs identified in this study. The slope 1:1 is indicated by a red line.