Complete mitochondrial genomes and updated divergence time of the two freshwater clupeids endemic to Lake Tanganyika (Africa) suggest intralacustrine speciation

Abstract

Background: The hydrogeological history of Lake Tanganyika paints a complex image of several colonization and adaptive radiation events. The initial basin was formed around 9-12 million years ago (MYA) from the predecessor of the Malagarasi-Congo River and only 5-6 MYA, its sub-basins fused to produce the clear, deep waters of today. Next to the well-known radiations of cichlid fishes, the lake also harbours a modest clade of only two clupeid species, Stolothrissa tanganicae and Limnothrissa miodon. They are members of Pellonulini, a tribe of clupeid fishes that mostly occur in freshwater and that colonized West and Central-Africa during a period of high sea levels during the Cenozoic. There is no consensus on the phylogenetic relationships between members of Pellonulini and the timing of the colonization of Lake Tanganyika by clupeids.

Results: We use short-read next generation sequencing of 10X Chromium libraries to sequence and assemble the full mitochondrial genomes of S. tanganicae and L. miodon. We then use Maximum likelihood and Bayesian inference to place them into the phylogeny of Pellonulini and other clupeiforms, taking advantage of all available full mitochondrial clupeiform genomes. We identify Potamothrissa obtusirostris as the closest living relative of the Tanganyika sardines and confirm paraphyly for Microthrissa. We estimate the divergence of the Tanganyika sardines around 3.64 MYA [95% CI: 0.99, 6.29], and from P. obtusirostris around 10.92 MYA [95% CI: 6.37-15.48].

Conclusions: These estimates imply that the ancestor of the Tanganyika sardines diverged from a riverine ancestor and entered the proto-lake Tanganyika around the time of its formation from the Malagarasi-Congo River, and diverged into the two extant species at the onset of deep clearwater conditions. Our results prompt a more thorough examination of the relationships within Pellonulini, and the new mitochondrial genomes provide an important resource for the future study of this tribe, e.g. as a reference for species identification, genetic diversity, and macroevolutionary studies.

Keywords: Clupeiformes; Great Lakes; Mitogenome; Phylogenetics; Time calibration.

Fig. 1

Mitochondrial genomes of Stolothrissa tanganicae (upper) and Limnothrissa miodon (lower). Inner circle shading indicates GC-content. Outer circle: black regions are protein coding genes, red regions are tRNA genes, beige regions are rRNA genes, and brown is the D-loop. Regions closer to the centre are located on the minus strand

Fig. 2

Nucleotide diversity (π) at mitochondrial PCGs (green) and rRNA genes (blue). π was calculated for 107 clupeiform and 6 non-clupeiform fish species using a sliding window with a window size of 150 bp and steps of 35 bp. Values could not be calculated for the gene coding for ND4, ND6, ND4L and for the D-loop (red)

Fig. 3

Outgroup-rooted maximum likelihood phylogeny of Clupeiformes. Topology and branch lengths were estimated based on mitochondrial protein-coding genes, rRNA genes and D-loop sequence of 107 clupeiform and 6 non-clupeiform fishes. Node support was assessed by Shimodaira-Hasegawa-like approximate likelihood ratio tests (SH-aLRT%) and ultrafast bootstrap (UFBoot%). Nodes with SH-aLRT% < 75 and UFBoot% < 90 were polytomized and their support values are not shown. The scale bar indicates model-corrected evolutionary distance (expected number of nucleotide substitutions per site). Subfamilies are indicated on the right side in black, families in colour. Pellonulini is highlighted in green

Fig. 4

Outgroup-rooted Bayesian phylogeny of Clupeiformes. Topology and branch lengths were estimated based on mitochondrial protein-coding genes, rRNA genes and D-loop sequence of 107 clupeiform and 6 non-clupeiform fishes. Node support was assessed by Bayesian posterior probabilities (BPP). Nodes with BPP < 0.85 were polytomized and their support values are not shown. Probabilities were rounded to the nearest 0.01. The scale bar indicates model-corrected evolutionary distance (expected number of nucleotide substitutions per site). Subfamilies are indicated on the right side in black, families in colour. Pellonulini is highlighted in green

Fig. 5

Outgroup-rooted time-calibrated phylogeny of Clupeiformes. Divergence times were estimated using BEAST, based on the first and second codon positions of 13 mitochondrial protein coding genes of 107 clupeiform and 6 non-clupeiform fishes. Blue bars represent Bayesian 95% credibility intervals. Calibration points (C1–C3) are indicated on the corresponding nodes. Pellonulini is highlighted in green