Mitochondrial DNA phylogeography of the Labeobarbus intermedius complex (Pisces, Cyprinidae) from Ethiopia

Journal of Fish Biology (2014) 85, 228–245 doi:10.1111/jfb.12408, available online at wileyonlinelibrary.com

Mitochondrial DNA phylogeography of the Labeobarbus intermedius complex (Pisces, Cyprinidae) from Ethiopia K. A. Beshera* and P. M. Harris Biodiversity and Systematics, Department of Biological Sciences, Box 870345, The University of Alabama, Tuscaloosa, AL 35487-0345, U.S.A. (Received 7 May 2013, Accepted 14 March 2014) Mitochondrial DNA phylogeography of populations of the Labeobarbus intermedius complex (hexaploid barb) was investigated using 88 complete and 71 partial cytochrome b (cytb) sequences originating from 21 localities in five major drainages in Ethiopia and two localities in northern Kenya. The samples included 14 of the 15 Labeobarbus species described from Lake Tana. Discrete phylogeographic analyses of 159 cytb sequences employing Bayesian Markov Chain Monte Carlo (MCMC) simulations using Bayesian evolutionary analysis by sampling trees (BEAST) supported the monophyly of the L. intermedius complex, including the Lake Tana species. This analysis, in combination with statistical parsimony analysis, identified two mitochondrial DNA lineages within the complex. Divergence dating employing coalescent simulations suggested that the geographic split in the L. intermedius complex that led to the formation of these lineages occurred during the Pleistocene (c. 0⋅5 M b.p.), consistent with the timing of volcano-tectonic events postulated to have shaped the current landscape of East Africa. © 2014 The Fisheries Society of the British Isles

Key words: East Africa; haplotypes; hexaploid minnow; lineages; Pleistocene; Rift Valley.

INTRODUCTION Labeobarbus intermedius (Rüppell 1836) is a large hexaploid barb from East Africa well known for its extensive morphological diversity (Banister, 1973). Since its description, the taxonomic limits of L. intermedius have not been well established, and there have been only two studies (morphology, Banister, 1973; molecular, de Graaf et al., 2010) with relatively wider geographic sampling that examined intraspecific variation, although this was not the focus of the latter study. The taxonomic uncertainty surrounding L. intermedius has prompted several authors (Nagelkerke & Sibbing, 1996; de Graaf et al., 2008, 2010) to refer to it as L. intermedius complex. The name Labeobarbus intermedius complex is used throughout this study. The type locality of L. intermedius complex is Lake Tana in the north western highlands of Ethiopia (Rüppell, 1836), but the redescription of the species by Banister (1973) was based on the examination of 454 specimens collected from lakes and rivers across Ethiopia and Lake Baringo, Kenya (0∘ 37′ N; 36∘ 04′ E). Thus far, no molecular studies compared the nominal taxon L. intermedius with the sympatric and allopatric *Author to whom correspondence should be addressed at present address: Wallace Community College, 1141 Wallace Dr., Dothan, AL 36303, U.S.A. Tel.: +1 3345562637; email: [email protected]

228 © 2014 The Fisheries Society of the British Isles

PHYLOGEOGRAPHY OF LABEOBARBUS INTERMEDIUS

229

Labeobarbus species. For most of its taxonomic history, the L. intermedius complex was classified in the genus Barbus. Recently, Skelton (2001, 2002) reclassified the large hexaploid barbs of Africa into Labeobarbus based on karyological (Golubstov & Krysanov, 1993) and phylogenetic (Machordom & Doadrio, 2001) evidence. Throughout this study, the taxonomic recommendation of Skelton (2001, 2002) is adopted. The existence of several morphologically distinct forms in Lake Tana led to multiple species descriptions in the early literature (e.g. six by Rüppell, 1836; 10 by Boulenger, 1902, 1907, 1911; 10 species and 23 sub-species by Bini, 1940). The earliest comprehensive taxonomic account of L. intermedius complex was based on Banister’s (1973) revision of the large barbs of East and Central Africa. In his revision, Banister (1973) synonymyzed >50 nominal species and sub-species in East Africa, including those from Lake Tana, into the L. intermedius complex (Golubstov et al., 2002). Banister (1973) also recognized two sub-species: Labeobarbus intermedius intermedius (Rüppell 1836), widely distributed throughout Ethiopia and northern Kenya and Labeobarbus intermedius australis (Banister 1973) found in Lake Baringo, Kenya. Several authors subsequently regarded Banister’s (1973) taxonomic conclusions as putative, at best. Skorepa (1992) described Banister’s (1973) sub-specific separation of L. i. australis as ‘unreal’, although he maintained Banister’s (1973) synonymy of several large barb species and sub-species with the L. intermedius complex. Nagelkerke & Sibbing (1997, 2000) questioned Banister’s (1973) hypothesis of a single species of large barb and described 14 species from Lake Tana distinct from the L. intermedius complex. Golubstov et al. (2002) suggested that populations of the L. intermedius complex from the Ethiopian Rift Valley represent multiple evolutionary lineages. The L. intermedius complex is widely distributed throughout Ethiopia and also enters into northern Kenya (Banister, 1973). In Ethiopia, the L. intermedius complex is found in all the major basins including the Rift Valley lakes, the Omo, Genale, Wabi Shebele, Awash and Abay (including Lake Tana) River basins (Banister, 1973; Roberts, 1975). This complex has also been reported from the Baro River system, a tributary of the White Nile River, within the boundaries of Ethiopia (Krysanov & Golubstov, 1996). In Kenya, the L. intermedius complex has been recorded from the Uasso Nyiro River and Lake Turkana (3∘ 30′ N; 36∘ 00′ E) basins (Skorepa, 1992), as well as from Lake Baringo (Banister, 1973). H Y D R O G R A P H I C F E AT U R E S O F E T H I O P I A

The combined effects of faulting, uplifting and volcanism during the Tertiary period were the primary geologic forces shaping the current hydrographic features within Ethiopia (Mohr, 1966). These tectonic processes formed the Ethiopian Rift Valley, the northernmost branch of the East African Rift System, creating a division between the highlands of the north-west and south-east (Fig. 1). Three distinct drainage patterns occur in the river basins of the north-western highlands. Most rivers lying on the northern and western escarpment of these highlands (Abay, Barbo and Tekeze River basins) flow to the west and form part of the Blue Nile River drainage. Rivers originating from the eastern or north-eastern escarpment of the highlands flow east into either the endoreic basins of the Rift Valley or Awash River. The headwaters of the Awash River are on the south-eastern side of the escarpment; this river flows north-east through the Danakil Depression into Lake Abe. Rivers on the southern side of the escarpment flow south into Lake Turkana via the Omo River system. Rivers originating from the south-eastern

© 2014 The Fisheries Society of the British Isles, Journal of Fish Biology 2014, 85, 228–245

230

K . A . B E S H E R A A N D P. M . H A R R I S

dS Re

Eritrea

ea

Yemen

Tekeze Basin

Abay Basin

G

s Rif

in

t Va lley

as eB

Lak e

el

South Sudan

13 18 14 15 16 17

GBL

eb Sh

9

Omo Basin

i ab W

GBL12

Og ad en

20

19 8

11

Ba sin

lia

7 10

6

So ma

Ethiopia

asin

20

ANL Barro-Akobo Basin

den

fA

o ulf

D

3 1 2 4 5

Aw ash B

Sudan

Jib ou ti

Danakil Basin Lake Tana

Genale-Dawa Basin

Lake Turkana

Uganda

Kenya

Fig. 1. Map showing major drainages in Ethiopia, sampling localities of Labeobarbus intermedius complex (with some points offset for clarity) and distribution of lineages ( , lineage A; , lineage B). Sampling site codes are indicated in Table I. The geographic position of sampling localities 19, 20 and 21 were approximated. ANL, Addis Ababa-Nekemt Lineament; GBL, Goba-Bonga Lineament.

highlands flow south-east (Genale, Dawa, Wabi Shebele and Fafan Rivers) into Somalia. The major hydrographic features of the Rift Valley are a chain of endoreic lakes. These lakes, together with the Awash River system, form three major basins within the Ethiopian Rift Valley, i.e. (1) the Awash River drainage in the north, (2) two systems of linked lakes in the central Rift Valley (Zeway-Langano-Abiyata-Shala and Awasa-Shallo) and waters flowing into and out of these lakes and (3) Lakes Abaya, Chamo and Chew Bahir and their tributaries in the south. It has been hypothesized that in the late Tertiary, these endoreic lakes formed a continuous system (Mohr, 1966) but during the pluvial period, which lasted from Late Glacial into early Holocene, only the four northern Rift lakes (Lakes Abiyata, Langano, Shala and Zeway) were unified into a larger lake, which drained to the north into the Awash River (Grove et al., 1975). During the same period, the Chew Bahir basin had an overflow into Lake Turkana, while the latter was connected with the White Nile system via the Baro-Akobo (Sobat) River basin (Grove et al., 1975; Grove, 1983). Lake Turkana occurs in the northern Kenyan Rift, which is connected with the southern Ethiopian Rift Valley by a 300 km wide zone of overlap (Ebinger et al., 2000). Despite its broad geographic distribution and potential value as a model organism to study the diversification of aquatic organisms in East Africa, phylogenetic relationships within the L. intermedius complex have not been fully clarified and very little is known about the geographic distribution of its genetic diversity. A previous phylogenetic analysis (de Graaf et al., 2010) of the complex was based on limited geographic sampling and, therefore, was insufficient to draw firm conclusions

© 2014 The Fisheries Society of the British Isles, Journal of Fish Biology 2014, 85, 228–245

PHYLOGEOGRAPHY OF LABEOBARBUS INTERMEDIUS

231

on its phylogeography suggesting the need for a comprehensive phylogeographic analysis. The aims of this study were to show the distribution of genetic variation within the L. intermedius complex across its geographic range based on the analysis of complete cytochrome b (cytb) gene sequences (1141 bp) and test if (1) extant pattern of genetic variation in the L. intermedius complex reflects histories of river drainages, (2) phylogeographic structure of populations corresponds to hypothesized previous basin connections summarized above and (3) phylogeographic structure of the L. intermedius complex corresponds to its taxonomy.

MATERIALS AND METHODS SAMPLING Tissue samples of the L. intermedius complex (including the species flock of Labeobarbus from Lake Tana) for this study were collected from 13 water bodies in four major drainages in Ethiopia between January and May 2006. Fishes were collected using gillnets, and assistance was provided by local fisheries experts in the identification of specimens. Additional tissue samples of the L. intermedius complex representing four water bodies were obtained from museum collections. All tissues were preserved in 95% ethanol. Specimens examined in this study, basin and water body sampled, sampling localities (for samples collected for this study), GenBank accession numbers and number of individuals examined for each water body are summarized in Table I. D NA E X T R AC T I O N A N D S E QU E N C I N G Total genomic DNA was extracted using the DNeasy tissue extraction kit (Qiagen; www.qiagen.com) following the manufacturer’s instructions. The mitochondrial (mt) cytb gene (1141 bp) was amplified using primers and PCR protocols developed by Briolay et al. (1998). Sequences were generated via dye terminator reactions and read on an ABI 3100 prism sequencer (www.lifetechnologies.com). Sequences were aligned independently by eye using Bioedit (Hall, 1999). Complete cytb sequences of 43 unique mtDNA haplotypes were deposited in GenBank and assigned accession numbers JN886992–JN887034 (Table I). P H Y L O G E O G R A P H I C A N A LY S I S The present data comprises 159 cytb sequences of the L. intermedius complex from 21 localities representing five drainages across Ethiopia and two localities in Kenya. Of these, 86 sequences (1141 bp) were generated for this study while the remaining 73 sequences [71 partial (GQ853201–GQ853271; de Graaf et al., 2010) and two complete (AF112406 and AF780872) cytb sequences] were retrieved from GenBank. Phylogeographic pattern within the L. intermedius complex was inferred based on cytb data employing discrete phylogeographic analysis, which integrates spatial, temporal and genealogy inference, under the Bayesian statistical framework implemented in Bayesian evolutionary analysis by sampling trees (BEAST; Lemey et al., 2009). In discrete phylogeographic analysis, no outgroup taxa were used. For estimation of time to most recent common ancestor (TMRCA), five sequences representing three taxa, Labeobarbus bynni bynni (Forskål 1775), Labeobarbus bynni occidentalis (Boulengar 1911) and Labeobarbus petitjeani (Daget 1962), were chosen based on the relationships generated by a preliminary phylogenetic analysis of a wide range of related taxa and used as outgroups. The BEAST input file was generated using BEAUTI 1.4.6 (Drummond & Rambaut, 2007). Later, necessary additions for Bayesian stochastic search variable selection (BSSVS) that enables discrete state reconstructions and inference of most parsimonious description of phylogeographic pattern (Lemey et al., 2009) were made using text editor. The geographic distributions of sampling locations for cytb sequences were incorporated into the input file as prior specifications.

© 2014 The Fisheries Society of the British Isles, Journal of Fish Biology 2014, 85, 228–245

4 4 4 2 4 3 9

2 4 4 4

Labeobarbus brevicephalus

Labeobarbus crassibarbis

Labeobarbus dainellii Labeobarbus gorgorensis

Labeobarbus gorguari

L. intermedius

Labeobarbus longissimus Labeobarbus macrophthalmus

Labeobarbus megastoma

Labeobarbus Negdia

4

5 30

NS

Labeobarbus acutirostris

L. intermedius

Gumara River

Lake Tana

Labeobarbus intermedius L. intermedius

Abay River Dedessa River

Abay

Species

Water body

Basin

11, 12, 29, 31

11, 28,

5, 11 12, 28, 29

4, 11, 13, 28

11, 28,

11, 28 12, 37, 40

12,

11, 29, 31

11, 28,

2, 10, 21, 12

1, 2, 3, 4 5–9, 32, 33, 34, 41–44

NH JN886992–JN886995 JN886996–JN887001, GQ53251–GQ53266 JN886993, JN887002–JN887004 GQ850201–GQ850203, JN887030 GQ853204, GQ853205, JN887033, JN887003 GQ853206–GQ853208, JN887004 GQ853209, GQ853210 GQ853211–GQ853213, JN887034 GQ853214, GQ853215, JN887029 GQ853236–GQ853238, JN887003, JN887005–JN887007 GQ853216, JN887006 GQ853217–GQ853219, JN887031 GQ853220–GQ853221, JN887033 GQ853223–GQ853225, JN887031

GenBank accession number

Table I. List of samples of Labeobarbus spp. examined in this study along with basin and water body sampled

2

2

1 2

1

1

3

1

1, 2

2

4

5 6

Site number (Fig. 1)

232 K . A . B E S H E R A A N D P. M . H A R R I S

© 2014 The Fisheries Society of the British Isles, Journal of Fish Biology 2014, 85, 228–245

Lake Langano Bulbula River Lake Abaya Lake Chamo

© 2014 The Fisheries Society of the British Isles, Journal of Fish Biology 2014, 85, 228–245

N S , number of specimens; N H , haplotype number.

Rift Valley

L. intermedius L. intermedius L. intermedius L. intermedius L. intermedius L. intermedius L. intermedius L. intermedius L. intermedius Labeobarbus intermedius australis Labeobarbus byinni bynni Labeobarbus bynni occidentalis Labeobarbus petitjeanii

L. intermedius L. intermedius L. intermedius L. intermedius

Awash River Borkena River Lake Awassa

6 3 5 2 4 5 3 1 1 1 1 1 1

5 1 2 8

2 4 4 5 1 8

Labeobarbus Surkis Labeobarbus truttiformis Labeobarbus tsanensis L. intermedius L. intermedius L. intermedius

Kulfo River ArbaMinch Springs Sagoe River Darse River Omo Gibe River Gilgel Gibe River Gojeb River Wabi Shebele Arer River Turkana Lake Kamnarok (Kenya) Lake Baringo Lake Baringo (Kenya) Outgroup taxa Nile River, Egypt Bafing River, Guinea Bafing River, Guinea

Rift Valley

5

Labeobarbus platydorsus

Awash

NS

Species

Water body

Basin

GenBank accession number

11, 29, 30, 39 GQ853226–GQ853228, JN887031, JN887032 11, 12 GQ853229, JN887003 11, 29, 38 GQ853230–GQ853232, JN887031 11, 12, 28 GQ853233–GQ853235, JN887003 16 JN887008, JN887009 16 GQ853271 16, 45 JN887008, JN887010, GQ853244–GQ853245 14, 15 JN887011–JN887013 21 GQ853246 25 JN887026, JN887028 25, 26 JN887026, JN887027, GQ853239–GQ853242 19, 20 JN887018, JN887019 16, 17, 18 JN887014–JN887017 19 JN887018 18, 19 JN887017, JN887018 17, 18 JN887016, JN887017, JN887020 22, 23, 24 JN887021–JN887023 17, 24 JN887024, JN887025 36 GQ853247 27 AF112406 4 AF180872 AF28742 AF180829, AF287421 AF287443, AF180875

NH

Table I. Continued

14 15 17 18 10 11 11 21

8 19 13 16

1 2 2 7 20 9

2, 3

Site number (Fig. 1)

PHYLOGEOGRAPHY OF LABEOBARBUS INTERMEDIUS

233

234

K . A . B E S H E R A A N D P. M . H A R R I S

The best performing evolutionary model was obtained based on the Akaike information criterion (AIC; Huelsenbeck & Crandall, 1997) using MODELTEST 3.7 (Posada & Crandall, 1998). Results from MODELTEST were applied to set model parameters for discrete phylogeographic analyses and estimation of TMRCA. Analyses were performed employing BEAST via the freely available BIOPORTAL platform (Kumar et al., 2009; www.bioportal.uio.no) with the GTR nucleotide substitution model, an uncorrelated lognormal relaxed molecular clock model, and a Bayesian coalescent model assuming constant population size. A generalized teleost cytb substitution rate of 0⋅76–2⋅20% per million years was used as a uniform prior (Berendzen et al., 2008) in all analyses. Because estimates of Labeobarbus-specific fossil dates are not available, no calibration points were used to root the trees. For discrete phylogeographic analysis, five independent Markov Chain Monte Carlo (MCMC) simulations each with 200 million generations were run sampling trees every 20 000th generations; in TMRCA estimation, 50 million generations were run sampling every 10 000th generations. Results from the five runs were combined using LogCombiner 1.6.2 (Rambaut & Drummond, 2011a). To confirm stationarity and determine the mean and 95% highest posterior density (HPD) for TMRCA and other statistics, outputs of the runs were viewed using Tracer 1.5 (Rambaut & Drummond, 2009). TreeAnotator 1.6.2 (Rambaut & Drummond, 2011b) and FigTree 1.3.1 (Rambaut, 2010) were used, respectively, to summarize the posterior tree distribution and visualize the annotated maximum clade credibility (MCC) tree. The number of unique haplotypes and haplotype (H d ± s.d.) and nucleotide (pi ) diversities were calculated in DNASP 4.5 (Rozas et al., 2003). Haplotype parsimony networks were constructed in TCS (Clement et al., 2005) using cytb haplotype dataset comprising 45 unique haplotype sequences employing statistical parsimony, a genetic algorithm introduced by Templeton et al. (1992). In this analysis, maximum numbers of mutational steps that make parsimonious connections between haplotype sequences were calculated with 95% confidence. A matrix of pair-wise uncorrected p-distances among haplotypes was generated using phylogenetic analysis using parsimony* (PAUP*) based on maximum likelihood settings generated in MODELTEST (Posada & Crandall, 1998). Genetic divergences between and within lineages were calculated based on this matrix.

RESULTS S E L E C T I O N O F M O D E L O F N U C L E OT I D E S U B S T I T U T I O N

MODELTEST 3.7 (Posada & Crandall, 1998), based on AIC, determined that the model that best described the evolution of the cytb sequence data in the L. intermedius complex was the general time reversible (GTR-I; −lnL = 2188⋅4771, AIC = 4394⋅9541, K = 9), proportion of invariant sites (pInv = 0⋅8322) and equal rates for all sites. PHYLOGEOGRAPHY

The MCC tree generated by discrete phylogeographic analyses (Fig. 2) strongly supports the monophyly of the L. intermedius complex (posterior probability, PP = 100%). An interesting discovery was that the L. intermedius complex was split into two mtDNA lineages, lineage A (PP = 58%) and lineage B (PP = 96%). Of the 159 sequences examined, 118 (74%) fell into lineage A while the remaining 41 sequences (26%) fell into lineage B. Lineage A consisted of sequences geographically originating from the Abay (Blue Nile) River including Lake Tana (localities 1–6), Awash River (localities 7 and 20), northern Rift Valley Basins (NRV; localities 8 and 9) and Lake Baringo (Kenya). Sequences that originated from Omo River (localities 10–12), southern Ethiopian Rift Valley (SRV; localities 13–18), Wabi Shebele River

© 2014 The Fisheries Society of the British Isles, Journal of Fish Biology 2014, 85, 228–245

235

PHYLOGEOGRAPHY OF LABEOBARBUS INTERMEDIUS

100

Dedessa River

100 Awash River and NRV lakes

Lineage A

58 Dedessa River

Lake Baringo

91

Lake Tana

100

Lake Tana

Lake Tana

Gilgelgibe River

Lineage B

96

Lakes Abaya and Chamo

100

Lake Kamnarok 80

100

80

60

40

20

Kulfo and Sagoe Rivers 0

Fig. 2. Maximum clade credibility tree obtained from Bayesian analysis of cytochrome b (cytb) sequences, showing two lineages (lineages A and B) and relationships among populations of Labeobarbus intermedius complex. Branch supports (posterior probability values) are indicated above branches. Scale shows branch length.

(locality 21), Bulbula River (locality 19) and Lake Turkana Basins comprise lineage B (Figs 1 and 2). An unexpected result was the placement of the lone sequence from Bulbula River within lineage B. The TCS analysis of mitochondrial data of the L. intermedius complex produced a single network with all haplotypes connected. In this haplotype network (Fig. 3), there were 23 mutational steps between the most distant haplotypes. The haplotype network consisted of two mtDNA groups (Fig. 1), consistent with the phylogeny generated from discrete phylogeographic analyses. The first group contained haplotypes originating from Abay River (haplotypes 1–13, 28–35 and 37–44), Awash River and NRV (haplotypes 14, 15, 35 and 45) drainages and Lake Baringo (haplotype 4, shared with Abay River drainage) while the second group contained haplotypes originating from SRV

© 2014 The Fisheries Society of the British Isles, Journal of Fish Biology 2014, 85, 228–245

236

K . A . B E S H E R A A N D P. M . H A R R I S

13

1 28 37

29

12

7

31 8

42

9

33

41

10

2 43

30

44

40

3

35

4

15

38

11

34 14 5 6 45

39

32 26 25 17

16 20

19 18

24 27

22 21 23 36

Fig. 3. Statistical parsimony network for cytochrome b (cytb) sequences of 45 unique Labeobarbus intermedius complex haplotypes connected to each other with 95% c.i. Circles correspond to unique haplotypes with numbers in circles indicating haplotype number. Circle sizes reflect the frequency of each haplotype. Lines connecting circles (including the small circles) correspond to mutation steps. Geographic origins of haplotypes are colour-coded ( , Abay River drainage; , northern Rift Valley-Awash River drainage; , Omo River drainage; , southern Rift Valley; , Wabi Shebele River drainage; , Lake Baringo (Kenya); , Turkana drainage (Kenya)).

(haplotypes 16–20, 25 and 26), Omo River (haplotypes 17 and 18 shared with SRV and 21–24), Wabi Shebele River (haplotype 36) and Lake Turkana (haplotype 27) basins. Intralineage relationships were largely unresolved with mostly poorly supported terminal nodes (PP < 50%). Nevertheless, some monophyletic groups were apparent within lineages A and B: (1) the group comprising populations from the Awash River and NRV lakes (PP = 100%), (2) the Lake Chamo + Lake Abaya group (PP = 100%)

© 2014 The Fisheries Society of the British Isles, Journal of Fish Biology 2014, 85, 228–245

PHYLOGEOGRAPHY OF LABEOBARBUS INTERMEDIUS

237

and (3) Kulfo River + Sagoe River group (PP = 80%). The first group fell within lineage A while the last two fell within lineage B. All sequences of the species flock of Labeobarbus from Lake Tana were placed within lineage A, but relationships among the species were not resolved. GENETIC DIVERSITY AND DIVERGENCE AND TMRCA

Mean sequence divergence between lineage A and lineage B, after excluding missing sites (due to partial cytb sequences retrieved from GenBank database), was 1⋅27% while it was 0⋅93% using the whole dataset. According to Bayesian analysis (BEAST), estimated divergence time between lineage A and lineage B was 0⋅5 million years before present (M b.p.) [95% credible interval (c.i.): 0⋅22–0⋅98 M b.p.]. The ages of the most recent common ancestors (TMRCAs) of lineages A and B were 0⋅24 M b.p. (95% c.i.: 0⋅11–0⋅98 M b.p.) and 0⋅15 M b.p. (95% c.i.: 0⋅05–0⋅31 M b.p.), respectively. A total of 45 unique mtDNA haplotypes were identified among 159 sequences of the L. intermedius complex: 32 haplotypes in lineage A (representing 118 sequences) and 13 haplotypes in lineage B (representing 41 sequences). Overall, a relatively high level of haplotype diversity (mean ± s.d. H d = 0⋅950 ± 0⋅007), but low level of nucleotide diversity (pi = 0⋅00695) was observed. Haplotype (H d ) and nucleotide (pi ) diversities within lineage A were 0⋅92 ± 0⋅012 and 0⋅0033 (mean ± s.d.), respectively, while they were 0⋅87 ± 0⋅032 and 0⋅0028, respectively, in lineage B. Within the L. intermedius complex, pair-wise haplotype sequence divergences (expressed as per cent p-distances) varied from 0⋅08 to 1⋅65% (Table II) and average within-lineage haplotype sequence divergences were 0⋅76% (lineage A) and 0⋅93% (lineage B).

DISCUSSION PHYLOGEOGRAPHY OF THE L. INTERMEDIUS COMPLEX

Discrete phylogeographic analysis of mtDNA sequence variation in the L. intermedius complex recovered two lineages: lineages A and B (Fig. 2). Lineage A was represented by sequences originating from Abay (Blue Nile) and Awash River drainages, NRV lakes and Lake Baringo (Kenya), whereas lineage B comprised sequences from Omo and Wabi Shebele River drainages, Lake Turkana basin and southern Ethiopian Rift Valley populations. The geographic distributions of the two lineages of the L. intermedius complex showed no overlap with the exception of haplotype 21, which represents sequences originating from the Bulbula River in the NRV and Omo River basin, suggesting historical isolation between the lineages. Haplotypes, however, were shared extensively among drainages and populations within the two lineages. This pattern is exemplified by haplotypes 17 and 18, which were dispersed throughout the Omo River and SRV drainages, while haplotype 14 was shared among populations of Awash River and Lakes Awassa and Langano. There is paucity of information on the phylogeographic patterns of other co-distributed fish species of the region. Consequently, it is not possible to determine whether the phylogeographic patterns within the L. intermedius complex reflect a common pattern of evolution in aquatic taxa or are unique to this species complex.

© 2014 The Fisheries Society of the British Isles, Journal of Fish Biology 2014, 85, 228–245

HAP1 HAP2 HAP3 HAP4 HAP5 HAP6 HAP7 HAP8 HAP9 HAP10 HAP11 HAP12 HAP13 HAP14 HAP15 HAP16 HAP17 HAP18 HAP19 HAP20 HAP21 HAP22 HAP23 HAP24 HAP25 HAP26 HAP27 HAP28 HAP29 HAP30 HAP31 HAP32 HAP33 HAP34

– 0⋅35 0⋅44 0⋅44 0⋅79 1⋅05 0⋅61 0⋅35 0⋅53 0⋅70 0⋅44 0⋅09 0⋅44 0⋅61 0⋅70 1⋅32 1⋅23 1⋅32 1⋅40 1⋅32 1⋅40 1⋅49 1⋅49 1⋅32 1⋅40 1⋅49 1⋅40 0⋅18 0⋅35 0⋅61 0⋅18 1⋅22 0⋅68 0⋅61

0⋅09 0⋅09 0⋅44 0⋅70 0⋅26 0⋅18 0⋅18 0⋅35 0⋅09 0⋅26 0⋅44 0⋅26 0⋅35 0⋅96 0⋅88 0⋅96 1⋅05 0⋅96 1⋅05 1⋅14 1⋅14 0⋅96 1⋅05 1⋅14 1⋅05 0⋅35 0⋅18 0⋅26 0⋅35 0⋅92 0⋅29 0⋅31

0⋅18 0⋅35 0⋅61 0⋅35 0⋅26 0⋅26 0⋅44 0⋅18 0⋅35 0⋅53 0⋅35 0⋅44 0⋅88 0⋅79 0⋅88 0⋅96 0⋅88 0⋅96 1⋅05 1⋅05 0⋅88 0⋅96 1⋅05 0⋅97 0⋅44 0⋅26 0⋅35 0⋅44 0⋅82 0⋅39 0⋅20

0⋅53 0⋅79 0⋅35 0⋅26 0⋅26 0⋅44 0⋅18 0⋅35 0⋅53 0⋅35 0⋅44 1⋅05 0⋅96 1⋅05 1⋅14 1⋅05 1⋅14 1⋅23 1⋅23 1⋅05 1⋅14 1⋅23 1⋅14 0⋅44 0⋅26 0⋅35 0⋅44 1⋅02 0⋅38 0⋅41

0⋅96 0⋅70 0⋅44 0⋅61 0⋅79 0⋅53 0⋅70 0⋅88 0⋅70 0⋅79 1⋅23 1⋅14 1⋅23 1⋅32 1⋅23 1⋅32 1⋅40 1⋅40 1⋅23 1⋅32 1⋅40 1⋅32 0⋅79 0⋅61 0⋅70 0⋅79 1⋅22 0⋅58 0⋅20

0⋅96 0⋅88 0⋅88 0⋅88 0⋅79 0⋅96 1⋅14 0⋅96 1⋅05 0⋅61 0⋅53 0⋅61 0⋅70 0⋅79 0⋅70 0⋅79 0⋅79 0⋅61 0⋅70 0⋅79 0⋅70 1⋅05 0⋅88 0⋅96 1⋅05 0⋅10 1⋅07 0⋅92 0⋅44 0⋅09 0⋅61 0⋅35 0⋅53 0⋅70 0⋅53 0⋅61 1⋅23 1⋅14 1⋅23 1⋅32 1⋅23 1⋅32 1⋅40 1⋅40 1⋅23 1⋅32 1⋅40 1⋅32 0⋅61 0⋅44 0⋅53 0⋅61 1⋅22 0⋅58 0⋅61 0⋅35 0⋅53 0⋅26 0⋅26 0⋅44 0⋅44 0⋅53 1⋅14 1⋅05 1⋅14 1⋅23 1⋅14 1⋅23 1⋅32 1⋅32 1⋅14 1⋅23 1⋅32 1⋅23 0⋅35 0⋅35 0⋅44 0⋅35 1⋅12 0⋅29 0⋅31 0⋅53 0⋅26 0⋅44 0⋅61 0⋅44 0⋅53 1⋅14 1⋅05 1⋅14 1⋅23 1⋅14 1⋅23 1⋅32 1⋅32 1⋅14 1⋅23 1⋅32 1⋅23 0⋅53 0⋅35 0⋅44 0⋅53 1⋅12 0⋅48 0⋅51 0⋅26 0⋅61 0⋅79 0⋅61 0⋅70 0⋅96 0⋅88 0⋅96 1⋅05 0⋅96 1⋅05 1⋅14 1⋅14 0⋅96 1⋅05 1⋅14 1⋅05 0⋅70 0⋅53 0⋅61 0⋅53 1⋅12 0⋅67 0⋅71 0⋅35 0⋅53 0⋅35 0⋅44 1⋅05 0⋅96 1⋅05 1⋅14 1⋅05 1⋅14 1⋅23 1⋅23 1⋅05 1⋅14 1⋅23 1⋅14 0⋅44 0⋅26 0⋅35 0⋅26 1⋅02 0⋅39 0⋅41 0⋅35 0⋅53 0⋅61 1⋅23 1⋅14 1⋅23 1⋅32 1⋅23 1⋅32 1⋅40 1⋅40 1⋅23 1⋅32 1⋅40 1⋅32 0⋅09 0⋅26 0⋅53 0⋅09 1⋅12 0⋅58 0⋅51 0⋅70 0⋅79 1⋅40 1⋅32 1⋅40 1⋅49 1⋅40 1⋅49 1⋅58 1⋅58 1⋅40 1⋅49 1⋅58 1⋅49 0⋅44 0⋅44 0⋅70 0⋅44 1⋅33 0⋅67 0⋅71 0⋅09 1⋅23 1⋅14 1⋅23 1⋅32 1⋅23 1⋅32 1⋅40 1⋅40 1⋅23 1⋅32 1⋅40 1⋅32 0⋅61 0⋅44 0⋅53 0⋅61 1⋅12 0⋅58 0⋅51 1⋅32 1⋅23 1⋅32 1⋅40 1⋅32 1⋅40 1⋅49 1⋅49 1⋅32 1⋅40 1⋅49 1⋅40 0⋅70 0⋅53 0⋅61 0⋅70 1⋅22 0⋅67 0⋅61

0⋅09 0⋅18 0⋅09 0⋅18 0⋅26 0⋅35 0⋅35 0⋅18 0⋅26 0⋅35 0⋅26 1⋅14 1⋅14 1⋅23 1⋅32 0⋅61 1⋅36 1⋅23

0⋅09 0⋅18 0⋅26 0⋅18 0⋅26 0⋅26 0⋅09 0⋅18 0⋅26 0⋅18 1⋅05 1⋅05 1⋅14 1⋅23 0⋅51 1⋅27 1⋅13

0⋅09 0⋅18 0⋅26 0⋅35 0⋅35 0⋅18 0⋅26 0⋅35 0⋅09 1⋅14 1⋅14 1⋅23 1⋅32 0⋅61 1⋅36 1⋅23

0⋅09 0⋅35 0⋅44 0⋅44 0⋅26 0⋅35 0⋅44 0⋅18 1⋅23 1⋅23 1⋅32 1⋅40 0⋅71 1⋅46 1⋅33

0⋅44 0⋅53 0⋅53 0⋅35 0⋅44 0⋅53 0⋅26 1⋅14 1⋅14 1⋅23 1⋅32 0⋅81 1⋅36 1⋅23

0⋅09 0⋅09 0⋅26 0⋅35 0⋅44 0⋅35 1⋅23 1⋅23 1⋅32 1⋅40 0⋅71 1⋅45 1⋅33

0⋅18 0⋅35 0⋅44 0⋅53 0⋅44 1⋅32 1⋅32 1⋅40 1⋅49 0⋅81 1⋅55 1⋅43

HAP1 HAP2 HAP3 HAP4 HAP5 HAP6 HAP7 HAP8 HAP9 HAP10 HAP11 HAP12 HAP13 HAP14 HAP15 HAP16 HAP17 HAP18 HAP19 HAP20 HAP21 HAP22

Table II. Pair-wise genetic distances (expressed as per cent uncorrected p-distances) in the mitochondrial cytochrome b gene between haplotypes of Labeobarbus intermedius complex

238 K . A . B E S H E R A A N D P. M . H A R R I S

© 2014 The Fisheries Society of the British Isles, Journal of Fish Biology 2014, 85, 228–245

HAP24 HAP25 HAP26 HAP27 HAP28 HAP29 HAP30 HAP31 HAP32 HAP33 HAP34 HAP35 HAP36 HAP37 HAP38 HAP39 HAP40 HAP41 HAP42 HAP43 HAP44 HAP45

HAP35 HAP36 HAP37 HAP38 HAP39 HAP40 HAP41 HAP42 HAP43 HAP44 HAP45

0⋅22 1⋅32 0⋅39 0⋅19 0⋅19 0⋅19 0⋅22 0⋅29 0⋅10 0⋅29 0⋅38

0⋅33 1⋅22 0⋅49 0⋅29 0⋅29 0⋅29 0⋅34 0⋅39 0⋅20 0⋅39 0⋅48

0⋅33 1⋅42 0⋅48 0⋅29 0⋅29 0⋅29 0⋅33 0⋅39 0⋅19 0⋅38 0⋅48

0⋅77 1⋅62 0⋅87 0⋅68 0⋅68 0⋅68 0⋅78 0⋅78 0⋅58 0⋅77 0⋅87

0⋅99 0⋅91 1⋅17 0⋅97 0⋅97 0⋅97 1⋅11 1⋅07 0⋅88 1⋅07 1⋅16

0⋅55 1⋅62 0⋅68 0⋅49 0⋅48 0⋅48 0⋅22 0⋅20 0⋅39 0⋅19 0⋅67

0⋅44 1⋅52 0⋅38 0⋅39 0⋅39 0⋅39 0⋅33 0⋅49 0⋅29 0⋅48 0⋅57

0⋅44 1⋅52 0⋅58 0⋅39 0⋅39 0⋅39 0⋅11 0⋅10 0⋅29 0⋅10 0⋅57

0⋅65 1⋅32 0⋅77 0⋅38 0⋅38 0⋅57 0⋅55 0⋅68 0⋅48 0⋅67 0⋅76

0⋅33 1⋅42 0⋅49 0⋅10 0⋅09 0⋅29 0⋅33 0⋅39 0⋅20 0⋅39 0⋅48

0⋅44 1⋅52 0⋅10 0⋅49 0⋅49 0⋅49 0⋅33 0⋅59 0⋅39 0⋅58 0⋅67

0⋅66 1⋅73 0⋅39 0⋅58 0⋅58 0⋅58 0⋅55 0⋅68 0⋅48 0⋅67 0⋅77

0⋅11 1⋅52 0⋅68 0⋅48 0⋅48 0⋅48 0⋅44 0⋅58 0⋅39 0⋅58 0⋅10

0⋅22 1⋅62 0⋅77 0⋅58 0⋅58 0⋅58 0⋅55 0⋅68 0⋅48 0⋅67 0⋅19

1⋅31 0⋅40 1⋅46 1⋅27 1⋅27 1⋅07 1⋅34 1⋅36 0⋅97 1⋅36 1⋅45

1⋅20 0⋅30 1⋅37 1⋅17 1⋅17 0⋅97 1⋅23 1⋅27 0⋅88 1⋅27 1⋅36

1⋅31 0⋅40 1⋅47 1⋅27 1⋅27 1⋅07 1⋅34 1⋅37 0⋅98 1⋅36 1⋅46

1⋅42 0⋅50 1⋅56 1⋅36 1⋅36 1⋅17 1⋅45 1⋅46 1⋅07 1⋅46 1⋅55

1⋅31 0⋅60 1⋅47 1⋅27 1⋅27 1⋅07 1⋅34 1⋅37 0⋅98 1⋅36 1⋅46

1⋅43 0⋅10 1⋅56 1⋅36 1⋅36 1⋅16 1⋅44 1⋅46 1⋅07 1⋅45 1⋅55

1⋅54 0⋅20 1⋅65 1⋅46 1⋅46 1⋅26 1⋅56 1⋅56 1⋅17 1⋅55 1⋅65

0⋅35 0⋅44 0⋅53 0⋅44 1⋅32 1⋅32 1⋅40 1⋅49 0⋅81 1⋅55 1⋅43 1⋅54 0⋅20 1⋅65 1⋅45 1⋅45 1⋅26 1⋅55 1⋅55 1⋅16 1⋅55 1⋅64

0⋅26 0⋅35 0⋅09 1⋅14 1⋅14 1⋅23 1⋅32 0⋅61 1⋅36 1⋅23 1⋅31 0⋅40 1⋅47 1⋅27 1⋅27 1⋅07 1⋅34 1⋅37 0⋅98 1⋅36 1⋅46

0⋅09 0⋅35 1⋅23 1⋅23 1⋅32 1⋅40 0⋅72 1⋅46 1⋅33 1⋅43 0⋅51 1⋅57 1⋅37 1⋅37 1⋅17 1⋅45 1⋅47 1⋅08 1⋅46 1⋅56

0⋅44 1⋅32 1⋅32 1⋅40 1⋅49 0⋅82 1⋅56 1⋅44 1⋅53 0⋅61 1⋅66 1⋅47 1⋅47 1⋅27 1⋅56 1⋅57 1⋅17 1⋅56 1⋅66

1⋅23 1⋅23 1⋅32 1⋅40 0⋅72 1⋅46 1⋅33 1⋅42 0⋅51 1⋅56 1⋅37 1⋅37 1⋅17 1⋅45 1⋅47 1⋅08 1⋅46 1⋅56

0⋅35 0⋅61 0⋅18 1⋅23 0⋅68 0⋅62 0⋅55 1⋅42 0⋅20 0⋅59 0⋅59 0⋅59 0⋅45 0⋅69 0⋅49 0⋅68 0⋅77 0⋅44 0⋅35 1⋅12 0⋅48 0⋅51 0⋅44 1⋅52 0⋅39 0⋅39 0⋅39 0⋅39 0⋅44 0⋅49 0⋅29 0⋅48 0⋅58 0⋅61 1⋅12 0⋅48 0⋅51 0⋅44 1⋅52 0⋅58 0⋅39 0⋅39 0⋅39 0⋅33 0⋅49 0⋅29 0⋅48 0⋅57 1⋅22 0⋅68 0⋅61 0⋅55 1⋅62 0⋅20 0⋅39 0⋅39 0⋅58 0⋅44 0⋅68 0⋅49 0⋅68 0⋅77 1⋅22 1⋅02 1⋅09 0⋅81 1⋅21 1⋅12 1⋅12 0⋅91 1⋅21 1⋅22 0⋅81 1⋅22 1⋅22 0⋅41 0⋅56 1⋅62 0⋅67 0⋅48 0⋅48 0⋅48 0⋅44 0⋅58 0⋅39 0⋅58 0⋅67 0⋅55 1⋅42 0⋅61 0⋅51 0⋅51 0⋅51 0⋅55 0⋅61 0⋅41 0⋅61 0⋅61

1⋅53 0⋅55 0⋅45 0⋅44 0⋅44 0⋅48 0⋅55 0⋅33 0⋅55 0⋅22

1⋅62 1⋅52 1⋅52 1⋅32 1⋅54 1⋅62 1⋅22 1⋅62 1⋅62

0⋅58 0⋅58 0⋅58 0⋅44 0⋅67 0⋅48 0⋅67 0⋅77

0⋅19 0⋅39 0⋅44 0⋅48 0⋅29 0⋅48 0⋅58

0⋅39 0⋅44 0⋅48 0⋅29 0⋅48 0⋅58

0⋅44 0⋅48 0⋅10 0⋅48 0⋅58

0⋅22 0⋅33 0⋅22 0⋅55

0⋅39 0⋅19 0⋅67

0⋅39 0⋅48

0⋅67

HAP23 HAP24 HAP25 HAP26 HAP27 HAP28 HAP29 HAP30 HAP31 HAP32 HAP33 HAP34 HAP35 HAP36 HAP37 HAP38 HAP39 HAP40 HAP41 HAP42 HAP43 HAP44

0⋅55 1⋅62 0⋅20 0⋅59 0⋅59 0⋅58 0⋅44 0⋅68 0⋅49 0⋅68 0⋅77

HAP1 HAP2 HAP3 HAP4 HAP5 HAP6 HAP7 HAP8 HAP9 HAP10 HAP11 HAP12 HAP13 HAP14 HAP15 HAP16 HAP17 HAP18 HAP19 HAP20 HAP21 HAP22

Table II. Continued

PHYLOGEOGRAPHY OF LABEOBARBUS INTERMEDIUS

© 2014 The Fisheries Society of the British Isles, Journal of Fish Biology 2014, 85, 228–245

239

240

K . A . B E S H E R A A N D P. M . H A R R I S

In the MCC tree obtained from discrete phylogeographic analysis, four sequences from Dedessa River (representing haplotypes 6 and 32) formed a clade sister to all other sequences in lineage A and had very weak support (PP = 58%). In the haplotype network, the two haplotypes were intermediate between the two lineages and more distant from haplotypes originating from the same river system as well as from populations in lineage A than they are from haplotypes in lineage B. For instance, haplotypes 6 and 32 were separated from haplotype 3 (their most closely related haplotype in lineage A) by seven and eight mutational steps, respectively, while the number of mutational steps between these haplotypes and haplotype 17 (their most closely related haplotype in lineage B) were six and five, respectively. These results may suggest the presence of undescribed biodiversity within Dedessa River, incomplete haplotype sampling or, given the complex geologic history of the region (Beadle, 1981) and the close proximity of the headwaters of these river systems, indicate post-divergence connections between Dedessa and Omo River systems. It is more likely that aspects of the current phylogeographic structure of the L. intermedius complex can be understood by considering historical geologic and climatic events of the region. According to Ebinger et al. (2000) and Bonnini et al. (2005), Africa experienced repeated paroxysms of volcano-tectonic activities during the Oligocene and the whole period following the Miocene. Divergence date estimates indicate that the two lineages of the L. intermedius complex diverged c. 0⋅5 M b.p. during the late Pleistocene. This suggests volcanic events that took place during the Pleistocene may have greatly altered the topography and drainage patterns of the region and thus may have significantly influenced genetic differentiation in the L. intermedius complex. The distinctness of the two mitochondrial DNA lineages of the L. intermedius complex with no shared haplotypes strongly suggests the existence of a long-term physical barrier to genetic exchange between drainages harbouring these lineages. Vicariant events associated with the formation of notable structural elements such as the Addis Ababa-Nekemt (ANL) and Goba-Bonga (GBL) tectonic lineaments may have affected genetic divergence in the L. intermedius complex. Within the Rift Valley system, haplotypes from localities north of the GBL fall within lineage A while those from localities south of the GBL fall within lineage B (Fig. 1). Similarly, haplotypes residing in localities on opposite sides of the ANL fall within different lineages. Volcanic activities that led to the formation of Addis Ababa-Nekemt Line (ANL) started at c.12-10 M b.p. and developed through three main phases: 12-7, 6-2 and