Available online at www.sciencedirect.com MOLECULAR SC IE■ N N C E C t E i Hll ^ I DIRECT® PHYLOGENETICS

EVOLUTION Molecular Phylogenetics and Evolution 34 (2005) 382-391 www.elsevier.com/locate/ympev

Mitochondrial phylogeny of the , a lineage of open-water endemic to , East Africa

Anita Brandstattera, Walter Salzburgerb, Christian Sturmbauerc’*

a Institute of Legal Medicine, Innsbruck Medical University, Mullerstr. 44, A-6020 Innsbruck, Austria b Lehrstuhl fur Zoologie und Evolutionsbiologie, Department of Biology, University of Konstanz, D-78457 Konstanz, Germany c Department o f Zoology, Karl-Franzens-University o f Graz, Universitatsplatz 2, A-8010 Graz, Austria

Received 25 June 2004; revised 25 October 2004

Abstract

We present a phylogeny of the Cyprichromini, a lineage of cichlid fishes from Lake Tanganyika, showing progressive adaptation towards pelagic life style. Our study is based upon three mitochondrial gene segments, 443 bp of the control region, 402 bp of the cytochrome b gene and the entire NADH dehydrogenase subunit 2 gene (1047 bp). The topologies obtained by different tree building methods subdivide the Cyprichromini into four distinct lineages: the -, the zonatus-, the Cyprichromis microlepidotus-lineage, and a lineage comprising Cyprichromis pavo and . Our study thus corroborates the distinctness of C. zonatus which was recently described formally. Concerning ecology and mating behavior, a clear evolutionary trend towards progressive adaptation to the pelagic zone emerges during the evolution of the Cyprichromini. The linearized tree analysis further shows that the four lineages have split almost contemporaneously. The mean Kimura-2-parameter distance among the four lineages emerging from the primary radiation of the Cyprichromini amounts to 7.21% and is in close agreement to that pre­ viously found for the primary radiation of the tribe Tropheini (7.01%), a lineage of rock-dwelling endemic to Lake Tangany­ ika. To date, the influence of lake level fluctuations as promoters of diversification has been demonstrated only for rock-dwelling cichlids. Based on the agreeement in temporary patterns of diversification, we suggest that Pleistocene lake level changes have left a similar genetic imprint in a group of cichlid fishes that progressively colonized the open water during their radiation. © 2004 Elsevier Inc. All rights reserved.

Keywords: Linearized tree; Adaptive radiation; Speciation; Mitochondrial DNA; Control region; Cytochrome b; NADH dehydrogenase subunit 2

1. Introduction several ancient lineages, which have radiated in parallel (Salzburger et al., 2002). These comprise both mouth- The three Great East African Lakes Victoria, brooders, which incubate their and fry in the buc­ Malawi and Tanganyika, each containing species cal cavity, and substrate breeders, which lay eggs on flocks of hundreds of endemic cichlid fishes, represent various substrates, and guard their offspring. Molecu­ important model systems for the study of explosive lar evidence suggested a polyphyletic origin of the Lake speciation, and adaptive radiation. While the species Tanganyika species flock and placed the tribe Cypri­ flock of Lake Malawi and the Lake Victoria “super­ chromini within the mouth-brooding “H-lineage” flock” are monophyletic and exclusively contain haplo- together with the Limnochromini, Perissodini, chromine species (Meyer et al., 1991; Verheyen et al., Ectodini, Haplochromini, and Tropheini (Kocher et al., 2003), the cichlid fauna of Lake Tanganyika contains 1995; Meyer et al., 1994; Nishida, 1991; Salzburger et al., 2002; Sturmbauer et al., 1994; but see Takahashi et * Corresponding author. Fax: +43 316 380 9875. al., 2001 who found several SINE markers that do not E-mail address: [email protected] (C. Sturmbauer). support monophyly of the “H-lineage”). The exact

1055-7903/$ - see front matter © 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.ympev.2004.10.019 A. Brandstatter et al. / Molecular Phylogenetics and Evolution 34 (2005) 382-391 383 placement of the Cyprichromini within the “H-lineage” some of which (e.g., Cyprichromis sp. “leptosoma is uncertain to date, though the most recent study goldfin” and Cyprichromis sp. “leptosoma jum bo-tri­ (Salzburger et al., 2002) placed the tribe as the sister color”) were included in this work. group of the tribes Limnochromini and Perissodini. Both genera are maternal mouth-brooders. The two The Cyprichromis was established by Scheuer­ species of the genus Paracyprichromis release and fer­ mann (1977) for four torpedo-shaped species, which tilize their eggs on a rocky surface, before the female previously were assigned to the genus Limnochromis. In picks them up to incubate them in the buccal cavity. the course of a revision of Lake Tanganyika cichlids, Males of C. leptosoma, C. microlepidotus, and C. zona- Poll (1986) divided the genus Cyprichromis according tus defend three-dimensional territories in the open to the number and arrangement of vertebrae into two water (the territories of C. zonatus measure a few genera, Cyprichromis sensu stricto and Paracyprich- meters diameter in horizontal and vertical direction omis, and established the tribe Cyprichromini. Cur­ and are situated above rocks in a lek-like arrangement; rently six species are recognized in the tribe: Takahashi et al., 2002) and females release their eggs in Paracyprichromis nigripinnis, Paracyprichromis brieni, the open water column to be fertilized inside the Cyprichromis leptosoma, Cyprichromis microlepidotus, female’s mouth, where they are subsequently incubated Cyprichromis pavo (Buscher, 1994; Poll, 1986) and the (Konings, 1998). In terms of spawning behavior recently described Cyprichromis zonatus, formerly C. pavo is clearly distinct from the other Cyprichromis called Cyprichromis sp. “zebra” in the aquarium trade species. It is not an open-water spawner but spawns (Takahashi et al., 2002). At least one more undescribed in close contact to the substrate (Konings, 1998), simi­ taxon (called Cyprichromis sp. “leptosoma jum bo” in lar to the two species of Paracyprichromis (Buscher, the aquarium trade) may exist. Species of the genus 1994). Cyprichromis feed on zoo- and phytoplankton and In terms of coloration males are generally more color­ occur gregariously off rocky slopes, often floating still ful than females, as in most other maternal mouth- with the head low. The species C. leptosoma and C. brooders. C. zonatus can be recognized by its overally microlepidotus follow diurnal movements of less conspicuous yellowish coloration and by broad ver­ and therefore may be observed near the surface. In con­ tical bars which are not permanently exhibited. Several trast, C. pavo predominantly occurs at depths of more distinctly colored geographic variants are recognized of than 20 m. C. zonatus usually occurs at depths below C. leptosoma, C. microlepidotus, and C. pavo. These spe­ 20 m and is often found in close proximity to C. pavo. cies also show polychromatism, i.e., differences in color Like Cyprichromis, Paracyprichromis feeds on zoo- and among males of the same population. Polychromatic phytoplankton in the open water column, but often males are not known from C. zonatus and from both above the rocky habitat. According to field observa­ Paracyprichromis species (Konings, 1998). Several tions several species co-occur in the same habitat. authors suggest that male coloration is a key trait for Buscher (1994), for example, found four species in female mate choice in cichlids promoting sexual selec­ sympatry at a 60 km stretch of coastline in the south­ tion (Deutsch, 1997; Dominey, 1984; Turner, 1994; western part of Lake Tanganyika: P. brieni, local vari­ Turner and Burrows, 1995). Sexual selection may thus ants of P. nigripinnis (“blue neon”) and C. leptosoma act as major driving force of (sympatric) speciation in (sp. “leptosoma jum bo”), as well as C. pavo. All mem­ some Cyprichromini, as in mouth-brooding haplochro- bers of the Cyprichromini form schools counting sev­ mines of Lake Victoria and Malawi, in which highly eral hundreds of individuals. Mixed species schools diverse male color patterns despite similar overall mor­ with representatives of C. leptosoma, C. zonatus, and P. phologies are observed among closely related species brieni are reported (Konings, 1998). (McKaye, 1991; Seehausen and van Alphen, 1998; Tay­ The two recognized species of Paracyprichromis occur lor et al., 1998). throughout the lake. C. microlepidotus and C. pavo have In this study, we analyzed the evolutionary a more or less complementary north/south distribution relationships within the tribe Cyprichromini on the with C. microlepidotus restricted to the northernmost basis of the molecular information of three mitochon­ part only and C. pavo in the central and southern waters drial gene segments. We also trace the evolutionary of Lake Tanganyika. C. leptosoma has a lake-wide distri­ transition from a more substrate-oriented (Paracy- bution except for the most northern parts (Buscher, prichromis, C. pavo, C. zonatus) to a more pelagic 1994; Konings, 1998), and C. zonatus only seems to lifestyle (C. leptosoma, C. microlepidotus). We obtained occur in the very south of the lake (Takahashi et al., samples of all described species and several distinctly 2002). Konings (1998) mentions a so far undescribed colored populations. We also addressed the succession species, Cyprichromis sp. “leptosoma jum bo” (also noted of major cladogenetic events in this group of fishes by Buscher, 1994) from the central lake region. In the using DNA sequences of the mitochondrial control aquarium trade, several geographic variants of C. lepto­ region, the most variable segment of the mitochondrial soma and Cyprichromis sp. “jum bo” have been imported, genome. 384 A. Brandstatter et al. / Molecular Phylogenetics and Evolution 34 (2005) 382-391

2. Materials and methods specimens are available from the authors and from the Lake Biwa Museum. 2.1. DNA sampling and extraction 2.2. Amplification and sequencing of mtDNA Eighty-three specimens belonging to the tribe Cypri- chromini from Lake Tanganyika were sequenced for as Total DNA was extracted from ethanol preserved fin- much as 1896bp of the mtDNA (1047bp of NADH clips using proteinase K digestion followed by sodium dehydrogenase subunit 2, ND2; 402 bp of cytochrome b, chloride extraction and ethanol precipitation (Bruford cyt b; and 447 bp of the control region, CR). Most of the et al., 1998). PCR-amplification and DNA-sequencing fishes were caught during two expeditions to Lake Tang­ were performed according to standard methods (Meyer anyika in 1999 and 2000; additional specimens were et al., 1990; Sturmbauer and Meyer, 1993) using an Air obtained from the Lake Biwa Museum (Japan) and from Thermo-Cycler (Idaho Technology) and the Big Dye the aquarium trade (see Table 1 and Fig. 1). Below we sequencing chemistry on an ABI 373A (Applied Biosys­ will use trade names for undescribed taxa. Two species tems). PCR was performed in a total volume of 17 ^l of the tribe Perissodini, Perissodus microlepis, and consisting of 1.7 ^l of 20 mM MgCl2+, 1.62 ^l Enzyme Perissodus straeleni were included as outgroup, since the Diluent (Idaho Technology), 0.085 ^l Taq polymerase tribe Perissodini was shown to be a sister group to the (GeneCraft), 1.7 ^l of each primer, 1.7 ^l dNTP-mix tribe Cyprichromini (Salzburger et al., 2002). Voucher (Idaho Technology), 6.8 ^l deionized water, and 1.7 ^l

Table 1 Numbers of samples analyzed of the Cyprichromini taxa studied Species Localitya DLPb CYTbb ND2b LT LBM AT Cyprichromis leptosoma Mpulungu (Zambia) 2 2 1 2 Cyprichromis leptosoma Wonzye point (Zambia) 1 1 1 1 Cyprichromis leptosoma “Malasa” Mpimbwe (Tanzania) 1 1 1 1 Cyprichromis leptosoma “Malasa” Tanzania 4 4 3 4 Cyprichromis leptosoma “Malasa” Utinta (Tanzania) 1 1 1 1 Cyprichromis leptosoma, blue caudal Wn Cape Kachese (Zambia) 10 13 2 13 Cyprichromis leptosoma, yellow caudal Wn Cape Kachese (Zambia) 16 17 2 17 Cyprichromis microlepidotus Bemba (Dem. Rep. Congo) 2 2 2 2 Cyprichromis microlepidotus Burundi 1 1 1 1 Cyprichromis microlepidotus Dem. Rep. Congo 1 1 1 Cyprichromis microlepidotus Kasai (Dem. Rep. Congo) 1 1 1 1 Cyprichromis microlepidotus Kigoma (Tanzania) 1 1 1 1 Cyprichromis pavo Katoto (Zambia) 1 1 1 1 Cyprichromis pavo Zambia 3 3 3 1 2 Cyprichromis sp“leptosoma goldfin” Chipimbi (Zambia) 2 2 1 2 Cyprichromis sp“leptosoma jumbo” Chisanza (Zambia) 2 2 1 2 Cyprichromis sp“leptosoma jumbo” Chituta Bay (Zambia) 1 1 1 1 Cyprichromis sp“leptosoma jumbo” Crocodile Island (Zambia) 1 1 1 1 Cyprichromis sp‘leptosoma jumbo” Kasenga (Zambia) 6 6 6 6 Cyprichromis sp“leptosoma jumbo” Katoto (Zambia) 3 3 3 3 Cyprichromis sp“leptosoma jumbo” Kitumba (Dem. Rep. Congo) 2 2 2 2 Cyprichromis sp“leptosoma jumbo” Tonga (Zambia) 1 1 1 1 Cyprichromis sp“leptosoma jumbo” Zambia 4 4 2 1 3 Cyprichromis sp“leptosoma jumbo-tricolor” Kambwimba (Tanzania) 1 1 1 1 Cyprichromis sp“leptosoma jumbo-tricolor” Mpimbwe (Tanzania) 1 1 1 1 Cyprichromis zonatus Kasenga (Zambia) 5 5 5 5 Cyprichromis zonatus Zambia 1 1 1 1 Paracyprichrom is brieni Chituta Bay (Zambia) 1 1 1 1 Paracyprichrom is brieni Crocodile Island (Zambia) 2 2 1 2 Paracyprichrom is brieni Zambia 1 1 1 1 Paracyprichrom s mgripmms Chituta Bay (Zambia) 2 2 1 2 Perissodus microlepis Tanzania 1 1 1 1 Perissodus straeleni Tanzania 1 1 1 1 LT, number of individuals caught in Lake Tanganyika during expeditions in 1999 and 2000. LBM, number of individuals provided by Lake Biwa Museum (Japan). AT, number of individuals obtained from the aquarium trade. a Coastal region where sample was obtained. b Number of individuals sequenced for the respective mitochondrial DNA region. A. Brandstatter et al. / Molecular Phylogenetics and Evolution 34 (2005) 382-391 385

the analyzes by one sequence only: This led to three final data sets comprising 41 different control region haplo­ types, 30 different cyt b haplotypes and 36 different ND2 haplotypes. After combining the three data sets (CR- ND2-Cytb-combined), sequence comparison led to the identification of 45 mitochondrial haplotypes. To test for the overall phylogenetic signal in the data set, a likeli­ hood mapping analysis (Strimmer and von Haeseler, 1997) was performed (10,000 quartets, random choice), using the computer program PUZZLE 4.0 (Strimmer and von Haeseler, 1996). For phylogenetic reconstruction we applied maxi- mum-likelihood (ML), maximum parsimony (MP), and neighbor-joining (NJ; (Saitou and Nei, 1987)) methods in parallel, using the computer program PAUP* 4.0b10 (Swofford, 2002). The robustness of the phylogenetic hypotheses was tested by bootstrapping (Felsenstein, 1985) and quartet-puzzling (Strimmer and von Haeseler, 1996). To quantify the substitution process and rate het­ erogeneity among sites, we used the computer program Fig. 1. Map of Lake Tanganyika, East Africa, with the location of its Modeltest Version 3.04 (Posada and Crandall, 1998). three deep sub-basins at a depth of 600 m, and the sample localities of Relevant Modeltest outputs are summarized in Table 2. Cyprichromini samples. MP topologies were obtained by heuristic search with 50 replicates using equal weighting (stepwise addition DNA template solution. The reaction cocktails were option), applying bootstrap analyzes with 1000 pseudo- heated to 94 °C (15 s) and then put through 38 reaction replicates as standard measure of confidence. In addition cycles: 94 °C for 0 s, 52 °C for 0 s, and 72 °C for 15 s. to searches based on equal weights of base substitutions, Primers used for both PCR and sequencing of cyto­ we performed MP analyzes, in which multiple hits were chrome b were L14724, 5' CGAAGCTTGATATGAA corrected by assigning weights according to the ML-esti- AAACCATCGTTC; and H15149, 5' AAACTGCAGC mated fourfold (ND2) or sixfold (cyt b) higher frequency CCCTCAGAATGATATTTGTCCTCA (Kocher et al., of transition (TS) over transversion-mutations (TV) in 1989). The first section of the mitochondrial control third codon positions. region was amplified and sequenced using the following To determine the sequence of the major cladogenetic primers: THR-2, 5' GCTT AC ACC AGT CTT GT AA events in the Cyprichromini, we constructed a linearized ACC; and TDK-D, 5' CCTGAAGTAGGAACCAGA tree based on a 365 bp segment of the D-Loop according TG (Kocher et al., 1989). For the amplification of ND2 to Baric et al. (2003), Sturmbauer et al. (2001; 2003), we used the primers MET, 5' CATACCCCAACAT Sturmbauer and Meyer (1992). We therefore performed GTTGGT; and TRP, 5' GAGATTTTCACTCCCGC the two-cluster test implemented in the computer pro­ TTA. For sequencing, we further applied ND2.2A, 5' gram LINTRE (Takezaki et al., 1995). First, rate con­ CTGACAAAAACTTGCCTT (Kocher et al., 1995). stancy was tested for all internal nodes in the topology based on the sequences of the control region. In our case, 2.3. Phylogenetic analyzes the test was performed applying the HKY + G (Hase- gawa et al., 1985) model of molecular evolution. No rate DNA sequences were aligned using the computer pro­ heterogeneity was detected at a high significance level gram Sequencher (GeneCodes, Ann Arbor, MI, USA). (pa < 0.01) so that none of the taxa had to be excluded for Identical haplotypes were identified and represented in further analyzes. Then, a tree for the given topology was

Table 2 Results of Modeltest and mean genetic differences Dataset Model selected —ln L Base frequencies I G Mean genetic diVerence 95% Confidence interval CR HKY + 1 + G 1386.16 A: 0.33; C: 0.21; G: 0.15; T: 0.33 0.61 0.69 5.74 (5.49-5.98) Cyt b HK Y + G 1029.34 A: 0.26; C: 0.32; G: 0.16; T: 0.27 0.00 0.18 5.61 (5.32-5.89) ND2 TIM + 1 + G 3289.99 A: 0.26; C: 0.35; G: 0.12; T: 0.27 0.58 1.55 5.17 (4.92-5.42) Combined TrN + 1 + G 5235.59 A: 0.28; C: 0.31; G: 0.13; T: 0.29 0.67 1.00 4.35 (4.21-4.49) I, proportion on invariable sites; G, gamma shape parameter. The mean genetic differences were computed based on the models of nucleotide substi­ tution suggested by Modeltest. 386 A. Brandstatter et al. / Molecular Phylogenetics and Evolution 34 (2005) 382-391 constructed under the assumption of rate constancy, Two major cladogenetic events became evident from which is termed a linearized tree. We used a previously the linearized tree analysis based on the control region- elaborated molecular clock calibration for East African data and HKY + G distances (Fig. 4). In the course of a lake cichlids (Sturmbauer et al., 2001), which was first cladogenetic event, four lineages emerged in the recently corroborated by additional data concerning the Cyprichromini, the Paracyprichromis-lineage, the C. split of the Lake Kivu and Lake Victoria haplochromine zonatus-lineage, the C. microlepidotus-lineage, and the C. fauna by the geological age of the Virunga Volcanoes leptosoma/pavo-lineage. The second cladogenetic event (Verheyen et al., 2003), to approximate absolute time resulted in the diversification of at least three of these scales for the major cladogenetic events. Our pairwise lineages. To compare the timing of the radiation events calculations were based upon the Kimura-2-parameter within the Cyprichromini with known diversification model (K2P), since this model was also used by events of the Tropheini (Sturmbauer et al., 2003) and Sturmbauer et al. (2001, 2003) and Baric et al. (2003). Tropheus (Baric et al., 2003; Sturmbauer and Meyer, 1992), mean Kimura-2-Parameter distances among clade members representing the two major diversification 3. Results events were calculated according to the branching pat­ terns of the linearized tree. An average K2P distance of The likelihood mapping analyzes demonstrated a 7.21% (95% CI, 7.17-7.24; number of pairwise compari­ strong phylogenetic signal in each data set: The ND2 sons, 1242) was found among the four major lineages data set yielded 97.2% fully resolved quartets, the cyt b arising from the first cladogenetic event, and of 2.25% data set 90.7% and the CR data set 93%. Within the (95% CI, 2.2-2.3) within subclades of the C. leptosoma- combined data set, likelihood mapping revealed a per­ C. pavo-lineage. centage fraction of 98.8% fully resolved quartet topolo­ gies, thus pointing to the increase of phylogenetic signal in proportion to the length of the analyzed DNA 4. Discussion sequences. Model corrected base frequencies and genetic differences are listed in Table 2. 4.1. Taxonomic and evolutionary implications The phylogenetic trees obtained by maximum likeli­ hood, maximum parsimony and neighbor joining were Our phylogenetic analysis consistently demonstrated very much alike. The genus Paracyprichromis was con­ the existence of four distinct lineages within the endemic sistently resolved as the most ancestral branch of tribe Tanganyikan tribe Cyprichromini, which were also con­ Cyprichromini. In all analyses of the combined data-set, firmed by the linearized tree analysis: the Paracyprichr- P. nigripinnis (two individuals with identical mtDNA omis lineage; the C. zonatus lineage; the C. microlepidotus sequences) was clearly distinguished from the three rep­ lineage, and the C. leptosoma/pavo lineage. The genus resentatives of P. brieni. In all resulting tree topologies, Paracyprichromis was unambiguously identified as the C. zonatus (six individuals) occupied the next ancestral most ancestral branch within the Cyprichromini. The branch. The next lineage was formed by six individuals next ancestral branch was formed by C. zonatus, thus of C. microlepidotus, followed by a clade containing all underscoring that C. zonatus is indeed a distinct species C. leptosoma, Cyprichromis sp. “leptosoma jum bo” and (Takahashi et al., 2002), because it was shown to repre­ C. pavo. The exact placement of C. pavo in relation to sent a primordial lineage, which never clustered with any subclades of C. leptosoma differed in the analyses. The other species. The six representatives of C. microlepidotus most inclusive analysis based upon the combined data covered a large portion of its distribution range and were set consistently placed C. pavo as sister group to a clade consistently resolved monophyletically. The C. leptosoma/ of nine individuals of C. leptosoma from the south-east­ pavo assemblage is taxonomically heterogeneous and not ern region of Lake Tanganyika (from Malasa Island, only comprises several distinctly colored populations of Tanzania, Mpulungu, Zambia), supported with high C. leptosoma and Cyprichromis sp. (“leptosoma jumbo,” bootstrap values (Fig. 2). These two taxa are sister group “leptosoma jumbo tricolor,” and “leptosoma goldfin”) to all other analyzed populations of C. leptosoma and but also C. pavo (see Fig. 3). The clustering of phenotypi­ Cyprichromis sp. “leptosoma jumbo.” In the neighbor cally distinct taxa and potentially new species (all joining tree based upon all DNA sequences of the con­ Cyprichromis sp. “leptosoma jumbo” are not formally trol region (Fig. 3), the clade “south” containing C. pavo described to date) suggests that this clade currently and C. leptosoma from the southeastern shore was undergoes speciation, possibly driven by sexual selection resolved within the clade of C. leptosoma and Cyprichr­ via mate choice, meriting further study by means of omis sp. “leptosoma jumbo.” Also, Cyprichromis sp. nuclear DNA markers. “leptosoma goldfin” and Cyprichromis sp. “leptosoma Concerning the evolutionary transition from a more jumbo-tricolor” were consistently placed within the C. substrate-oriented life-style to a truly pelagic life and leptosoma/pavo clade. spawning behavior, it seems interesting that the two A. Brandstatter et al. / Molecular Phylogenetics and Evolution 34 (2005) 382-391 387

Fig. 2. Maximum likelihood tree (substitution model: TrN + I + G; I = 0.67; G = 1.00) derived from the combined data set (1896bp). Score: 6127.20768. Outgroup taxa for the phylogenetic analyses were Perissodus straeleni and Perissodus microlepis. The top values on the branches corre­ spond to the results of the quartet puzzling analysis (Strimmer and von Haeseler, 1996), and the middle values refer to a MP bootstrap analysis (1000 pseudo-replicates) in which TV were weighted six times more than TS in cytochrome b and three times more in ND2 in third positions of all codons. The bottom values correspond to the neighbor-joining bootstrap values. most ancestral branches are the most substrate-oriented clade” of C. leptosoma. It is clearly distinct from the taxa. Albeit all Cyprichromini feed in the open water, other members of the C. leptosoma lineage, in that it they never venture far from the rocks. Male individuals spawns—like Paracyprichromis—in close contact to the of Paracyprichromis defend mating territories at rocks substrate, mostly along-side the vertical face of a rock or in caves, to where females are guided to . C. (Konings, 1998). To us, the most likely explanation is a zonatus males also maintain territories above a large reversal to a more substrate-oriented mode of spawning rock or bolder, with a diameter of several meters in hori­ in C. pavo, possibly as a consequence of its movement to zontal and vertical direction, but seem to already spawn greater water depths. in the open water. C. microlepidotus and C. leptosoma live and spawn in the open water, so that a clear evolu­ 4.2. Genetic traces o f lake level changes tionary trend towards progressing adaptation to the pelagic zone emerges. The only exception is C. pavo Two major bursts of cladogenesis become evident which was placed as the sister group to the “southern from our linearized tree analysis (Fig. 4). In the course 388 A. Brandstatter et al. / Molecular Phylogenetics and Evolution 34 (2005) 382-391

C. leptosoma. yellow caudal fin, Cape Kachese (Zambia) C. leptosoma, blue caudal fin, Cape Kachese (Zambia) nr C. leptosoma, yellow caudal. . fin, - Cape - Kachese (Zambia) C leptosoma, yellow caudal fin, Cape Kachese (Zambia) CO G leptosoma, yellow caudal fin, Cape Kachese (Zambia) C leptosoma, yellow caudal fin, Cape Kachese (Zambia) m “ G leptosoma, yellow caudal fin, Cape Kachese (Zamb ' C leptosoma, blue caudal fin. Cap«e Kachese (Zambia) . , x G leptosoma, yellow caudal fin, Cape Kachese (Zambia) C leptosoma, yellow caudal fin, Cape Kachese (Zambia) <5 C leptosoma, yellow caudal fin, Cape Kachese (Zambia) o C leptosoma, blue caudal fin, Cape Kachese (Zambia) C leptosoma, blue caudal fin, Cape Kachese (Zambia) 5 < C leptosoma, blue caudal fin. Cape Kachese (Zambia) o C leptosoma, yellow caudal fin, Cape Kachese (Zambia) C. leptosoma, yellow caudal fin, Cape Kachese (Zambia) - G leptosoma, blue caudal fin, Cape Kachese (Zambia) 0 -| " G leptosoma, blue caudal fin. Cape Kachese (Zambia) o LU C. leptosoma, blue caudal fin, Cape Kachese (Zambia) '«*» C. leptosoma, yellow caudal fin, Cape Kachese (Zambia) Q. C. leptosoma. blue caudal fin, Cape Kachese (Zambia) a 14 C. leptosoma, yellow caudal nn, Cape Kachese (Zambia) < C. leptosoma, yellow caudal fin, Cape Kachese (Zambia) C. leptosoma, yellow caudal fin, Cape Kachese (Zambia) C. leptosoma, yellow caudal fin, Cape Kachese (Zambia) o ■ C sp. “leptosoma jum bo”, Katoto (Zambia) P C sp. “leptosoma iumbo”, Katoto (Zambia) I— G leptosoma, blue caudal fin, Cape Kachese (Zambia) i* C. sp. “leptosoma jum bo”, Chisanza (Zambia) V) *- C sp, “leptosoma jum bo”, Kasenga (Zambia) — C. sp. “leptosoma jum bo”, Chisanza (Zambia) C sp. “leptosoma jum bo”, Kasenga (Zambia) C sp. “ leptosoma jum bo”, Kasenga (Zambia) £ “jumbo’ C sp. “ leptosoma jum bo”, Crocodile Island (Zambia) C sp. “ leptosoma jum bo”, Kasenga (Zambia) o G sp. “leptosoma jum bo”, Kasenga (Zambia) v. C. sp. “leptosoma jum bo”, Kasenga (Zambia) C. sp. “leptosoma jum bo”, Katoto (Zambia) >C 18 C. sp. “leptosoma jum bo”, Tonga (Zambia) C. sp. “leptosoma goldfin” Chifrimbi (Zambia) o C. sp. “leptosoma gold fin”, Chipimbi (Zambia) ‘goldfin’ C. leptosoma “Malasa” (Tanzania) 9 5 C. leptosoma “Malasa” (Tanzania) C. leptosoma ‘Malasa” (Tanzania) C. leptosoma “Malasa” (Tanzania) a 68 G leptosoma “Malasa”, Mpimbwe (Tanzania) 68 leptosoma, Mpulungu (Zambia) * “south” G leptosoma, Wonzye Point (Zambia) C. leptosoma, Mpulungu (Zambia) u “ “ C. leptosoma •'Malasa”, Utinta (Tanzania) I Cyprichromispa\>oCyprichromispavo (Zambia) 32 Cyprichromis pavo (Zambia) Cyprichromis I Cyprichromis pavo (Zambia) 78 *— Cyprichromis pavo, Kasenga (2(Zambia) pavo C. sp. “leptosoma jumbo-tricolor ”, Kambwimba (Tanzania) 1 C. sp. “leptosoma jumbo-tricolor ”, Mpimbwe (Tanzania) “jumbojrjcolpr G sp. “leptosoma jum bo (Zambia) G sp. “leptosoma jum bo”, Kitumba (Dem. Rep. Congo) G sp. “leptosoma jum bo” (Zambia) C. sp. “leptosoma jum bo”, Kitumba (Dem. Rep. Congo) “ jumbo” 43 C. sp. “leptosoma jum bo”, ChitutaBay (Zambia) C. sp. “leptosoma jum bo” (Zambia) C. sp. “leptosoma jum bo” (Zambia) I C. microlepidotus, Bemba (Zambia) C. microlepidotus (Dem. Rep. Congo) C. nicrolepidotus (Burundi) Cyprichromis 85 99 l - f— t C. microlepidotus, Kigoma (Tanzania) \ 1 C. milmicrolepidotus , Bemba (Zambia) microlepidotus C. microlepidotus,tn/‘ - Kasai (Zaire) C. zonatus; Kasenga (Zambia) C zonatus, Kasenga (Zambia) C. zonatus, Kasenga (Zambia) Cyprichromis 100 99 G zonatus, Kasenga (Zambia) [1 G zonatus, Kasenga (Zambia) zonatus £ G zonatus (Zambia) I Paracyprichrontis nigripinnis, Chituta Bay (Zambia) 1 Paracyprichromis nigripinnis, Chituta Bay (Zambia) Paracyprichromis brieni, Crocodile Island (Zambia) so r i “ Paracyprichromis brieni, Crocodile Island (Zambia) Paracyprichromis — - — *- Paracvprichromisp, brieni, Chituta Bay (Zambia) ParacyprichromisParacv, brieni (Zambia) i "■ Perissodusstraeleni (Tanzania) '------Perissodus microlepis (Tanzania) — 0.005 substitutions/site

Fig. 3. Neighbor-joining tree (substitution model HKY + I + G; I = 0.61; G = 0.69) derived from a 447 bp segment of the mitochondrial control region of 83 taxa, declaring Perissodus straeleni and Perissodus microlepis as outgroup. Numbers above the branches are corresponding bootstrap values (1000 pseudo-replicates). of the first cladogenetic event the four main lineages Lake Tanganyika has a complex geological history, within the Cyprichromini were established. The four characterized by extended periods of markedly lower major lineages arising from the primary radiation show lake level (Cohen et al., 1997; Lezzar et al., 1996). Sev­ an average K2P distance of 7.21% (95% CI, 7.17-7.24). eral lineages of Lake Tanganyika cichlids diversified in The second event concerns further diversification step with geology- and climate-induced changes of the within the C. leptosoma-C. pavo-lineage, with sub- lake habitat (Koblmuller et al., 2004; Salzburger et al., clades showing on average a K2P distance of 2.25% 2002). The radiation of the H-lineage (Salzburger et al., (95% CI, 2.2-2.3). 2002) and that of the Ectodini was suggested to be A. Brandstatter et al. / Molecular Phylogenetics and Evolution 34 (2005) 382-391 389

Fig. 4. Linearized tree based on a 365 bp segment of the control region. The linearized tree was compiled with the computer program LINTRE (Takezaki et al., 1995) after performing a relative rate test (Takezaki et al., 1995). The observed average K2P distance of 7.21% (95% CI, 7.17-7.24) between the four lineages in the Cyprichromini was used to date the cladogenetic events based on the molecular clock calibration of (Sturmbauer et al., 2001). The distance values on the top of the phylogram correspond to HKY + G distances, those on the bottom to K2P distances. 390 A. Brandstatter et al. / Molecular Phylogenetics and Evolution 34 (2005) 382-391 connected to the onset of deepwater conditions References (Koblmuller et al., 2004), dated to 5-6 million years ago (Tiercelin and Mondeguer, 1991). Lake level Xuctua- Baric, S., Salzburger, W., Sturmbauer, C., 2003. Phylogeography and tions were shown to have a great inXuence on population evolution of the Tanganyikan cichlid genus Tropheus based upon mitochondrial DNA sequences. J. Mol. Evol. 56, 54-68. subdivision and the speciation process of rock-dwelling Bruford, M., Hanotte, O., Brookfield, J., Burke, T., 1998. Multilocus cichlids such as the Tropheini and Eretmodini (Ruber and single-locus DNA Wngerprinting. In: Hoelzl, A. (Ed.), Molecu­ et al., 1998, 2001; Sturmbauer et al., 2001, 2003; Stur­ lar Genetic Analysis of Populations: a Practical Approach. Oxford mbauer and Meyer, 1992; Sturmbauer et al., 1997; University Press, London, pp. 287-336. Verheyen et al., 1996). In the tribe Tropheini, Stur­ Buscher, H., 1994. Cyprichromis pavo, ein neuer Cichlide aus dem Tang- anyika-See. Deutsche Aquarien- und Terrarienzeitschrift 4, 257-263. mbauer et al. (2003) found a K2P distance 7.01% (95% Cohen, A., Lezzar, K., Tiercelin, J., Soreghan, M., 1997. New palaeoge- CI 6.87-7.14) for the primary radiation of the Trophe- ographic and lake-level reconstructions of Lake Tangayika: impli­ ini, which is in close agreement to our Wnding of 7.21% cations for tectonic, climatic and biological evolution in a rift lake. for the primary radiation of the Cyprichromini. We Basin Res. 9, 107-132. recognize that the application of a molecular clock for Deutsch, J., 1997. Color variation in Malawi cichlids: evidence for adap­ tation, reinforcement or sexual selection?. Biol. J. Linn. Soc. 62, 1-14. estimating divergence times is problematic for several Dominey, W., 1984. Effects of sexual selection and life-history on speci- reasons (see e.g., Gillespie, 1991; Page and Holmes, ation: species Xocks in African cichlids and Hawaiian Drosophila. 1998). In particular, the calibration of the molecular In: Echelle, A., KornWeld, I. (Eds.), Evolution of Species clock for East African cichlid fishes remains difficult Flocks. University of Maine at Orono Press, Orono, pp. 231-249. due to the absence of a reliable fossil record. Consider­ Felsenstein, J., 1985. Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39, 783-791. ing the previously published molecular clock rate for Gillespie, J.H., 1991. The Causes of Molecular Evolution. Oxford Uni­ East African Lake cichlids (Sturmbauer et al., 2001), versity Press, Oxford. which was recently corroborated by Verheyen et al. Hasegawa, M., Kishino, H., Yano, T., 1985. Dating of the hum an-ape (2003), the maximum estimate for Wrst cladogenetic splitting by a molecular clock of mitochondrial DNA. J. Mol. Evol. event within the Cyprichromini would approximate 1.1 22, 160-174. Koblmuller, S., Salzburger, W., Sturmbauer, C., 2004. Evolutionary million years ago. Since the second diversiWcation event relationships in the sand-dwelling cichlid lineage of Lake concerning the C. leptosoma/C. pavo lineage is based Tanganyika suggest multiple colonization of rocky habitats and upon a geographically restricted array of taxa, further convergent evolution of biparental mouthbrooding. J. Mol. Evol. samples are needed to correlate this diversiWcation 58, 79-96. event to that of other lineages. However, the striking Kocher, T.D., Conroy, J.A., McKaye, K.R., StauVer, J.R., Lockwood, S.F., 1995. Evolution of N A D H dehydrogenase subunit 2 in east congruence of the primary radiation of the Tropheini African cichlid Wsh. Mol. Phylogenet. Evol. 4, 420-432. and Cyprichromini substantiates our hypothesis that a Kocher, T., Thomas, W., Meyer, A., Edwards, S., Paabo, S., Villa- major change in the lake environment triggered both blanca, F., Wilson, A., 1989. Dynamics of mitochondrial DNA evo­ radiations. Geological data in fact suggest that Lake lution in : ampliWcation and sequencing with conserved Tanganyika started to rise its lake level after a pro­ primers. Proc. Natl. Acad. Sci. USA 86, 6196-6200. Konings, A., 1998. Tanganyika Cichlids in Their Natural Habitat. longed period of low lake level during the late Pleisto­ Lezzar, K., Tiercelin, J., De Batist, M., Cohen, A., Bandora, T., Van cene (see Sturmbauer et al., 2001, 2003). Given that the Rensbergen, P., Le Turdu, C., Mifundu, W., Klerkx, J., 1996. New Cyprichromini inhabit the shallow pelagic zones of seismic stratigraphy and late tertiary history of the North Tangany­ Lake Tanganyika, and are therefore bound to the rock ika basin, East African rift system, deduced from multichannel and substrate to a much lesser extent than the strictly rock- high-piston core evidence. Basin Res. 8, 1-28. McKaye, K., 1991. Sexual selection and the evolution of the cichlid dwelling tribe Tropheini, it is remarkable to Wnd an Wshes of Lake Malawi, Africa. In: Keenleyside, M. (Ed.), Cichlid equal impact of lake level changes in a group of cichlid Fishes: Behaviour, Ecology, and Evolution. Chapman and Hall, Wshes that progressively colonized the open water dur­ New York, pp. 241-257. ing their radiation. Meyer, A., Kocher, T.D., Wilson, A.C., 1991. African Wshes. Nature 350, 467-468. Meyer, A., Montero, C., Spreinat, A., 1994. Evolutionary history of the cichlid species Xocks of the East African great lakes inferred from Acknowledgments molecular phylogenetic data. Adv. Limnol. 44, 407-423. Meyer, A., Kocher, T.D., Basasibwaki, P., Wilson, A.C., 1990. Mono- We thank T. Takahashi, P. Henninger, D. Kozak, and phyletic origin of Lake Victoria cichlid Wshes suggested by mito­ L. Onder who provided specimens, and the team at the chondrial DNA sequences. Nature 347, 550-553. Nishida, M., 1991. Lake Tanganyika as an evolutionary reservoir of Fisheries Department in Mpulungu, Zambia, for their old lineages of East African Wshes: inferences from allozyme data. help during Weld work. We are further grateful to J. Experientia 47, 974-979. Snoeks and T. Takahashi for species identiWcation, to Bob Page, R.D.M., Holmes, E.C., 1998. Molecular Evolution: A Phyloge­ Schelly and Sanja Baric for valuable comments on the netic Approach. Blackwell Sciences, Oxford. manuscript. W.S. was supported by the Austrian Acad­ Poll, M., 1986. Classification des Cichlidae du lac Tanganika. Tribus, genres et especes. Academie Royale de Belgique, Memoires de la emy of Sciences and the European Union, C.S. by the classe des sciences, Collection in -8°-2° serie, T. XLV-Fascicule, 2, Austrian Science Foundation (Project 12339 and 15239). 1-163. A. Brandstatter et al. / Molecular Phylogenetics and Evolution 34 (2005) 382-391 391

Posada, D., Crandall, K.A., 1998. MODELTEST: testing the model of Sturmbauer, C., Verheyen, E., Meyer, A., 1994. M itochondrial phylog- DNA substitution. Bioinformatics 14, 817-818. eny of the Lamprologini, the major substrate spawning lineage of Ruber, L., Verheyen, E., Sturmbauer, C., Meyer, A., 1998. Lake level cichlid fishes from Lake Tanganyika in eastern Africa. Mol. Biol. fluctuations and speciation in rock-dwelling cichlid fish in Lake Evol. 11, 691-703. Tanganyika, East Africa. In: Grant, P. (Ed.), Evolution on Islands. Sturmbauer, C., Verheyen, E., Ruber, L., Meyer, A., 1997. Phylogeo- Oxford University Press, London, pp. 225-240. graphic patterns in populations of cichlid fishes from rocky habi­ Ruber, L., Meyer, A., Sturmbauer, C., Verheyen, E., 2001. Population tats in Lake Tanganyika. In: Kocher, T., Stepien, C. (Eds.), structure in two sympatric species of the Lake Tanganyika cichlid Molecular Phylogeny of Fishes. Academic Press, San Diego, pp. tribe Eretmodini: evidence for introgression. Mol. Ecol. 10, 1207­ 97-111. 1225. Swofford, D., 2002. PAUP*: phylogenetic analysis using parsimony Saitou, N., Nei, M., 1987. The neighbor-joining method: a new m ethod (* and other methods), version 4.0.b10. Sinauer Associates, Sunder­ for reconstructing phylogenetic trees. Mol. Biol. Evol. 4, 406-425. land, MA. Salzburger, W., Meyer, A., Baric, S., Verheyen, E., Sturmbauer, C., Takahashi, K., Terai, Y., Nishida, M., Okada, N., 2001. Phylogenetic 2002. Phylogeny of the Lake Tanganyika cichlid species flock and relationships and ancient incomplete lineage sorting among cichlid its relationship to the central and East african haplochromine cich- fishes in Lake Tanganyika as revealed by analysis of insertion of lid fish faunas. Syst. Biol. 51, 113-135. retrotransposons. Mol. Biol. Evol. 18, 2057-2066. Scheuermann, H., 1977. A partial Revision of the genus Limnochromis Takahashi, T., Hori, M., Nakaya, K., 2002. New Species of Cyprichr- Regan 1920. Cichlidae (British Cichlid Association) volume III 2, omis (Perciformes: Cichlidae) from Lake Tanganyika, Afrika. 69-73. Copeia 4, 1029-1036. Seehausen, O., van Alphen, J.J., 1998. The effect of male coloration on Takezaki, N., Rzhetsky, A., Nei, M., 1995. Phylogenetic test of the female mate choice in closely related Lake Victoria cichlids (Hap- molecular clock and linearized trees. Mol. Biol. Evol. 12, 823-833. lochromis nyererei complex). Behav. Ecol. Sociobiol. 42, 1-8. Taylor, M., Turner, G., Robinson, T., Stauffer, J., 1998. Sexual selec­ Strimmer, K., von Haeseler, A., 1996. Quartet puzzling: a quartet maxi- tion, parasites and asymmetry of an extended phenotypic trait in a mum-likelihood method for reconstruction tree topologies. Mol. bower-building cichlid from Lake Malawi, Africa. Anim. Behav. 56, Biol. Evol. 13, 964-969. 379-384. Strimmer, K., von Haeseler, A., 1997. Likelihood-mapping. A simple Tiercelin, J.-J., Mondeguer, A., 1991. The geology of the Tanganyika method to visualize phylogenetic content of a sequence alignment. trough. In: Coulter, G.W. (Ed.), Lake Tanganyika and its Life. Proc. Natl. Acad. Sci. USA 94, 6815-6819. Oxford University Press, London, pp. 7-48. Sturmbauer, C., Baric, S., Salzburger, W., Ruber, L., Verheyen, E., 2001. Turner, G., 1994. Speciation mechanisms in Lake Malawi cichlids: a Lake level fluctuations synchronize genetic divergences of cichlid critical review. In: Martens, K., Goddeeris, B., Coulter, G. (Eds.), fishes in African lakes. Mol. Biol. Evol. 18, 144—154. Speciation in Ancient Lakes. Schweizerbart, Stuttgart, pp. 139-160. Sturmbauer, C., Hainz, U., Baric, S., Verheyen, E., Salzburger, W., Turner, G., Burrows, M., 1995. A model of sympatric speciation by sex­ 2003. Evolution of the tribe Tropheini from Lake Tanganyika: syn­ ual selection. Proc. R. Soc. Lond. B Biol. Sci. 260, 287-292. chronized explosive speciation producing multiple evolutionary Verheyen, E., Ruber, L., Snoeks, J., Meyer, A., 1996. M itochondrial parallelism. Hydrobiologia 500, 51-64. phylogeography of rock-dwelling cichlid fishes reveals evolu­ Sturmbauer, C., Meyer, A., 1992. Genetic divergence, speciation and tionary influence of historical lake level fluctuations of Lake morphological stasis in a lineage of African cichlid fishes. Nature Tanganyika. Philos. Trans. R. Soc. Lond. B Biol. Sci. 351, 797­ 358, 578-581. 805. Sturmbauer, C., Meyer, A., 1993. M itochondrial phylogeny of the Verheyen, E., Salzburger, W., Snoeks, J., Meyer, A., 2003. Origin of the endemic mouth-brooding lineages of cichlid fishes from Lake superflock of cichlid fishes from Lake Victoria, East Africa. Science Tanganyika in eastern Africa. Mol. Biol. Evol. 10, 751-768. 300, 325-329.