Molecular Phylogenetics and Evolution 38 (2006) 426–438

www.elsevier.com/locate/ympev

Phylogenetic relationships of the lamprologine cichlid genus Lepidiolamprologus (Teleostei: Perciformes) based on mitochondrial

and nuclear sequences, suggesting introgressive hybridization

Robert Schelly a,b,¤, Walter Salzburger c, Stephan Koblmüller d, Nina Duftner d,Christian Sturmbauer d

a Division of Vertebrate Zoology (Ichthyology), American Museum of Natural History, New York, NY 10024, USAb Department of Ecology, Evolution, and Environmental Biology, Center for Environmental Research and Conservation, Columbia University,

New York, NY 10027, USAc Lehrstuhl fuer Zoologie und Evolutionsbiologie, Department of Biology and Center for Junior Research Fellows, University of Konstanz,

D-78457 Konstanz, Germanyd Department of Zoology, Karl-Franzens-University of Graz, Universitätsplatz 2, A-8010 Graz, Austria

Received 21 April 2005; accepted 27 April 2005Available online 16 June 2005

Abstract

Using sequences of the mitochondrial NADH dehydrogenase subunit 2 gene (ND2, 1047 bp) and a segment of the non-codingmitochondrial control region, as well as nuclear sequences including two introns from the S7 ribosomal protein and the lociTmoM25, TmoM27, and UME002, we explore the phylogenetic relationships of Lepidiolamprologus, one of seven lamprologine cich-lid genera in Lake Tanganyika, East Africa. Analyses consisted of direct optimization using POY, including a parsimony sensitivityanalysis, and maximum likelihood and Bayesian inference for comparison. With respect to Lepidiolamprologus, the results based onthe mitochondrial dataset were robust to parameter variation in POY. Lepidiolamprologus cunningtoni was resolved in a large cladesister to ossiWed group lamprologines, among which the remaining Lepidiolamprologus were nested. In addition to L. attenuatus,L. elongatus, L. kendalli, and L. profundicola, Neolamprologus meeli, N. hecqui, N. boulengeri, N. variostigma, and two undescribedspecies were resolved in a two-pore Lepidiolamprologus clade sister to Lamprologus callipterus and two species of Altolamprologus.Lepidiolamprologus nkambae, in marked conXict with morphological and nuclear DNA evidence, nested outside of the two-pore Lep-idiolamprologus clade, suggesting that the mtDNA signal has been convoluted by introgressive hybridization. 2005 Elsevier Inc. All rights reserved.

Keywords: Cichlidae; Lamprologini; Lepidiolamprologus; Lake Tanganyika; Introgressive hybridization

1. Introduction

Among the 12 cichlid tribes recognized by Poll (1986)in Lake Tanganyika, East Africa, the substrate-broodinglamprologines are the most diverse, with about 80 spe-cies. Additionally, eight lamprologine species are foundin the Congo River (Schelly and Stiassny, 2004), and at

* Corresponding author. Fax: +1 212 769 5642.E-mail address: schelly@amnh.org (R. Schelly).

1055-7903/$ - see front matter 2005 Elsevier Inc. All rights reserved.doi:10.1016/j.ympev.2005.04.023

least one species occurs in the Malagarasi River (De Voset al., 2001; Schelly et al., 2003). While the monophyly ofPoll’s tribe Lamprologini has withstood scrutiny (Salz-burger et al., 2002a; Stiassny, 1997; Sturmbauer et al.,1994; Takahashi et al., 1998), most genera within thetribe are unquestionably polyphyletic. For instance,members of the “ossiWed group,” identiWed by Stiassny(1997) and distinguished by a labial bone suspendedwithin the labial ligament, are scattered among four ofseven lamprologine genera potentially rendering

R. Schelly et al. / Molecular Phylogenetics and Evolution 38 (2006) 426–438 427

Lamprologus, Neolamprologus, and Lepidiolamprologusnon-monophyletic.

Pellegrin (1904) originally erected the genus Lepidio-lamprologus for Lamprologus elongatus, deWning thenew genus, closely allied with Lamprologus, as some-what more elongate, with teeth-like Lamprologus;rather long gill rakers (12); small ctenoid scales num-bering 90–95 in longitudinal series; 18 dorsal spines;and 5 anal spines. Boulenger (1915) synonymized Lep-idiolamprologus with Lamprologus, and arrayed lamp-rologines in the genera Lamprologus, Julidochromis,and Telmatochromis, with no statement as to theirbeing part of a natural group. Subsequently, Regan(1920) recognized aYnities between the lamprologinesknown at the time, Telmatochromis, Julidochromis, andLamprologus, based on their strong conical teeth and4–10 anal spines. Regan (1920, 1922) argued that thediversity of Lamprologus species in the lake impliedthat the group originated in Lake Tanganyika, despitethe existence of Congo River representatives, which hebelieved were a single lineage. The Wrst signiWcant eVortto use osteology to guide lamprologine classiWcationwas that of Colombe and Allgayer (1985). In thatstudy, the genus Lamprologus was subdivided into Wvegenera based on characters of the infraorbital series,with only the Congo River species retained in the genusLamprologus. Pellegrin’s genus Lepidiolamprologus wasrehabilitated for six species (L. attenuatus, L. cunning-toni, L. elongatus, L. kendalli, L. nkambae, and L. pro-fundicola), and three new genera were created: themonotypic Variabilichromis for V. moorii, the mono-typic Paleolamprologus for P. toae, and Neolamprolo-gus for 38 species.

Poll (1986) retained the resurrected Lepidiolamprolo-gus, but criticized the suYciency of Pellegrin’s originalcharacters for the genus. Instead, he listed 61–73 lateralline scales, vs. 30–40 in other genera, plus a unique struc-ture of pelvic Wn rays and numerous scales in the occipi-tal, thoracic, and abdominal regions as supporting thegroup. Poll (1986) criticized the infraorbital charactersof Colombe and Allgayer because of their variabilitywithin species and even individuals. On these grounds,he altered their generic allocation in his new classiWca-tion. In addition to re-assigning several lake endemics tothe genus Lamprologus, Poll (1986) rejected the mono-typic genera Variabilichromis and Paleolamprologus, andadditionally proposed Altolamprologus as a new genus,for the highly distinctive A. compressiceps and A. calvus.Finally, Poll accepted Neolamprologus for most remain-ing Lake Tanganyika lamprologine species, with thecaveat that Neolamprologus would likely be further par-titioned in the future.

The most thorough morphology-based treatment oflamprologines was carried out by Stiassny (1997), wholisted a suite of osteological characters supporting lamp-rologine monophyly, in accord with numerous molecu-

lar studies (e.g., Salzburger et al., 2002a; Sturmbaueret al., 1994; Thompson et al., 1994). Unlike Poll (1986),Stiassny (1997) supported the creation of the genus Vari-abilichromis for V. moorii. Regarding the genus Lepidio-lamprologus, she suggested that L. cunningtoni should beexcluded, and N. pleuromaculatus, N. boulengeri, N. hec-qui, N. meeli, and N. lemairii should be included to ren-der the genus monophyletic. Stiassny highlighted theinadequacy of current lamprologine classiWcation bydeWning an “ossiWed group” of lamprologines, with rep-resentatives scattered among the genera Lamprologus,Neolamprologus, Lepidiolamprologus, and Altolamprolo-gus. OssiWed group lamprologines posses a sesamoidbone within the labial ligament, a condition mirrored incertain atherinomorphs, but unique among cichlids andperhaps even Perciformes. More recently, Takahashi(2003) used morphological characters to examine rela-tionships among Tanganyikan cichlids, but did notrecover the ossiWed group as a monophyletic assemblagein his lamprologine clade, consisting of 10 species only.

Utilizing two mtDNA loci and Wve nuclear loci for asubset of taxa, this study focuses on resolution of thephylogenetic relationships of species assigned to thegenus Lepidiolamprologus, one of the most distinctivegenera of the ossiWed group of lamprologines. ThemtDNA phylogeny is then used to trace the evolution oftwo distinctive morphological characters. We followPoll’s (1996) classiWcation, in which lamprologines com-prise seven genera: Altolamprologus Poll, 1986; Chali-nochromis Poll, 1974; Julidochromis Boulenger, 1898,Lamprologus Schilthuis, 1891, Lepidiolamprologus Pelle-grin, 1904; Neolamprologus Colombe and Allgayer,1985; and Telmatochromis Boulenger, 1898.

2. Materials and methods

2.1. Taxon sampling

In addition to 36 lamprologines, we included threeeretmodines (Spathodus erythrodon, Tanganicodus irsa-cae, and Eretmodus cyanostictus) and one perissodine(Perissodus microlepis), representatives of lineagesnested close to lamprologines in the analysis of Salz-burger et al. (2002a), as outgroups. Since the focus ofthis study was the genus Lepidiolamprologus, weincluded all but one species that has ever been placed inthat genus or suggested to be closely allied with it (onlyN. pleuromaculatus was unavailable), and two unde-scribed species, one with a Xank pigmentation patternsimilar to that of L. profundicola, fresh material ofwhich was collected in Zambia in March, 2004, and theother morphologically similar to N. boulengeri andN. meeli, collected in Zambia in October, 2001, andMarch, 2003. In addition, we thoroughly sampled fromthe ossiWed group of Stiassny (1997), including 20 out

428 R. Schelly et al. / Molecular Phylogenetics and Evolution 38 (2006) 426–438

of 26 members.1 The diversity of non-ossiWed lamprol-ogines was similarly well sampled, including a CongoRiver species [the newly described Lamprologus teug-elsi, previously identiWed as L. mocquardi in Salzburgeret al. (2002a) and Sturmbauer et al. (1994)], and represen-tative members of Neolamprologus, Julidochromis, andTelmatochromis. Among them were N. toae and N. moorii,for which monotypic genera have been proposed.

2.2. Molecular biological methods

Approximately, 1400 bp of mtDNA from 40 species(58 specimens in total) and 2700 bp of nuclear DNAfrom 18 species (26 specimens total) were sequenced, andare available from GenBank under Accession Nos.DQ054907–DQ055127. Voucher specimens have beendeposited at the Department of Zoology, University ofGraz, Austria, the Musée Royal de l’Afrique Centrale,the South African Institute for Aquatic Biodiversity, andthe American Museum of Natural History. Total DNAwas extracted from Wn clips or muscle tissue preserved in95% ethanol by using the Chelex 100 method (Walsh etal., 1991) or the Qiagen Tissue Extraction Kit followingthe manufacturer’s protocol. Polymerase chain reaction(PCR) was used to amplify portions of the mitochon-drial control region (»360 bp) and NADH dehydroge-nase subunit 2 gene (ND2, 1047 bp), as well as thenuclear loci UME002 (»405 bp), TmoM25 (»320 bp),TmoM27 (»365 bp), and the Wrst (570 bp) and second(»1085 bp) introns of the S7 ribosomal protein.

For the mitochondrial loci, reactions of 17�l totalvolume consisted of 6.5�l of deionized water, 1.7 �l ofdNTP mix, 1.7�l of 20 mM Mg2+ buVer, 1.7 �l of eachprimer, 1.62 �l of enzyme diluent, 0.085�l of Taq poly-merase, and 2�l of DNA extract. Primers used toamplify the control region were L-PRO-F, 5� AACTCTCACCCCTAGCTCCCAAAG (Meyer et al., 1994) andTDK-D, 5� CCTGAAGTAGGAACCAGATG (Kocheret al., 1989). For ND2, the primers used for ampliWcationwere MET, 5� CATACCCCAACATGTTGGT (Kocheret al., 1995) and TRP, 5� GAGATTTTCACTCCCGCTTA (Kocher et al., 1995). AmpliWcations were per-formed on an Air Thermo-Cycler (Idaho Technology)and consisted of a total of 40 cycles, beginning with 15 sat 94 °C for denaturation, followed by Wve cycles of 0 s at94 °C, 5 s at 48 °C for annealing, and 25 s at 72 °C forextension, followed by 35 cycles of 0 s at 94 °C, 0 s at52 °C, and 25 s at 72 °C. PCR products were puriWedusing the ExoSAP-IT kit (Amersham Biosciences), andboth strands were sequenced using the original primers,

1 One of our de facto ossiWed group members is Neolamprologusvariostigma, a species not examined by Stiassny (1997). Owing to theunavailability of material, we have been unable to observe the labialligament in any cleared and stained N. variostigma, but tentatively con-sider the labial bone to be present.

plus the ND2 internal primers ND2.2A, 5� CTGACAAAAACTTGCCTT (Kocher et al., 1995), and the newlydesigned primer ND2.Bob, 5� CTGGCAAAAACTTGCCCCTTT, with Big Dye Terminator Reaction Mix(Applied Biosystems). Sequencing reactions were elec-trophoresed on an ABI 373A automated sequencer(Applied Biosystems).

The Wve nuclear loci were ampliWed using tetrad ther-mocyclers (MJ Research) in reactions of 25�l totalvolume using one Ready-To-Go PCR bead (AmershamBiosciences), 2�l of DNA extract, and 1.25�l of eachprimer, of which the following were used: UME002f, 5�TCAGAGTGCAATGAGACATGAAT and UME002r,5� AATTTAGAAGCAGAAAATTAGACG (Parker andKornWeld, 1996); TmoM25f, 5� CTGCAGTGGCACATCAAGAATGAGCAGCGGT, TmoM25r, 5� CAAGAACCTTTCAAGTCATTTTG, TmoM27f, 5� AGGCAGGCAATTACCTTGATGTT, TmoM27r, 5� TACTAACTCTGAAAGAACCTGTGAT (Zardoya et al., 1996);S7RPEX1f, 5� TGGCCTCTTCCTTGGCCGT C, S7RPEX2r, 5� AACTCGTCTGGCTTTTCGCC, S7R PEX2f,5� AGCGCCAAAATAGTGAAGCC, and S7R PEX3r,5� GCCTTCAGGTCAGAGTTCAT (Chow and Haz-ama, 1998). AmpliWcations of the UME002 and S7 lociconsisted of 35 cycles, beginning with 6 min at 94 °C forinitial denaturation, followed by cycles of 60 s at 94 °C,60 s at 54–66 °C for annealing, and 1 min at 72 °C forextension, with a Wnal 6 min extension at 72 °C. AmpliW-

cations of the Tmo loci consisted of 39 cycles, beginningwith 5 min at 94 °C for initial denaturation, followed bycycles of 15 s at 94 °C, 5 s at 48 °C for annealing, and 30 sat 68 °C for extension, with a Wnal 7 min extension at72 °C. PCR products were puriWed using AMPure(Agencourt) and cycle-sequenced using Big Dye Termi-nator Reaction Mix (Applied Biosystems). Sequencingreactions were puriWed using CleanSEQ (Agencourt)and electrophoresed on an ABI 3730£l automatedsequencer (Applied Biosystems).

2.3. Phylogenetic analyses—direct optimization

Initially, parsimony was used to analyze a combineddataset of control region and ND2 sequences, withSpathodus erythrodon designated as the root, justiWed bythe analysis of Salzburger et al. (2002a). Further parsi-mony analyses were performed on a combined dataset ofthe Wve nuclear loci for a subset of the taxa in the mito-chondrial dataset, as well as a combined dataset with allseven nuclear and mitochondrial loci including all taxa.These analyses were performed using direct optimization(Wheeler, 1996) in POY (Wheeler et al., 2002) on theparallel computing cluster at the American Museum ofNatural History. Direct optimization, which treats inser-tions and deletions (indels) as transformation eventsalong with transitions (ts) and transversions (tv),optimizes raw sequences on topologies without recourse

R. Schelly et al. / Molecular Phylogenetics and Evolution 38 (2006) 426–438 429

to a multiple sequence alignment, thus avoiding the pos-sibility of tree search being adversely inXuenced by asub-optimal static alignment.

To speed up analyses in POY (Giannini and Sim-mons, 2003; Giribet, 2001; Wheeler, 2003a), datasets foreach locus were submitted as separate input Wles, withinwhich sequences were broken into fragments of 100–150 bp at invariant motifs. Initially, an expanded datasetcomprising 58 terminals (units of analysis), includingmultiple representatives of 10 species (2–4 individualsper species, including eight individuals for which controlregion sequences were unavailable), was analyzed undera gap:tv:ts cost regime of 1:1:1. The search strategybegan with 50 random addition sequences followed bytree bisection reconnection (TBR) branch swapping and20 rounds of TBR-ratchet (Nixon, 1999) in which 40%of the characters were reweighted by a multiplier of threein each round. In addition, tree fusing (GoloboV, 1999b)was performed on all replicates, allowing up to 250fusings and exchanging subtrees with no fewer thanthree taxa. Inputting the most parsimonious trees gener-ated by this stage, a Wnal reWnement was undertakenwith reduced iterations of ratcheting and tree fusing, butimplementing three dimensional optimization alignmentwith iterative pass (Wheeler, 2003a) and performing acomplete SankoV optimization on the downpass with thecommand–exact. Though computationally more inten-sive, this step was intended to increase the accuracy ofPOY heuristics. WinClada (Nixon, 2002) and NONA(GoloboV, 1999a) were used to verify tree lengths basedon implied alignments (Wheeler, 2003b), and to calculatethe consistency index, or CI (Kluge and Farris, 1969),and retention index, or RI (Farris, 1989).

Using the same strategy, additional datasets wereanalyzed: (1) comprising the mitochondrial loci and lim-ited to single representatives of each species (40 termi-nals), (2) comprising the nuclear loci for one or morerepresentatives of 18 species (26 terminals), and (3) com-prising all of the nuclear and mitochondrial loci for allterminals. For analyses 1 and 2, branch supports(Bremer, 1988, 1995) were calculated with TreeRot(Sorenson, 1996), and jackknife values were calculated inPAUP* 4.0b10 (SwoVord, 2002) by randomly deleting36.79% of the characters (Farris et al., 1996) for each of10,000 replications.

2.4. Sensitivity analysis

To explore the robustness of the topology derivedunder equal weights in relation to cost variation of gaps,transversions, and transitions, a sensitivity analysis wasperformed on the 40 terminal mitochondrial dataset(Giribet and Ribera, 2000; Wheeler, 1995). In addition tothe analysis under a 1:1:1 weighting scheme, a series ofanalyses under assorted diVerential weighting schemeswere performed. A total of 12 diVerent weighting

schemes were employed to test the eVects of downweigh-ting transversions, downweighting transitions, andexcluding transitions entirely, varying gap costs acrosseach scheme. The ratio of tv:ts was variously set to 1:1,1:2, 2:1, and 4:0, and for each of these schemes threeanalyses were run, with gaps weighted at 1, 2, and 4 timesthe larger of the tv:ts values (see Fig. 2, bottom-left). Thetwo-stage tree search approach outlined above was alsoused for each of these searches, with command-molecu-larmatrix specifying a text Wle with the appropriate step-matrix.

2.5. Maximum likelihood and Bayesian analyses

An explicit model of evolution was used for maximumlikelihood (ML) and Bayesian inference analyses forcomparison with the results obtained using parsimony. Inthese analyses, based on the 40 terminal mitochondrialdataset, all three eretmodines (not just Spathodus erythr-odon) were designated as the outgroup. Unlike for thePOY analyses, a static multiple sequence alignment wasrequired for the control region sequences of variablelengths. This was initially generated with Clustal W(Thompson et al., 1994) and manually reWned. To choosethe most appropriate model of sequence evolution forML and Bayesian inference, hierarchical likelihood ratiotest statistics were calculated using ModelTest 3.06(Posada and Crandall, 1998). The model recovered was(TrN + I +�) with nucleotide frequencies A D 0.28980,C D 0.30690, G D 0.12000, TD 0.28330, gamma shapeparameter (�) 0.9111, proportion of invariable sites0.4087, number of substitution types three, and R-matrixA M C, A M T, C M G, and GM T D 1.0000, A MG10.6289, and C MT 4.8090. The ML tree was calculatedwith heuristic searches consisting of 50 random additionsequence replicates in PAUP*, and as measures of conW-

dence, jackknife values (43 replicates; 36.79% characterdeletion) were calculated. A Bayesian inference tree wascalculated with MrBayes v3.0b4 (Huelsenbeck and Ron-quist, 2001). Posterior probabilities were obtained from a2£ 106 generation Metropolis-coupled Markov chainMonte Carlo simulation (10 parallel chains; chain tem-perature 0.2; trees sampled every 100 generations), withparameters estimated from the data set. We applied aburn-in of 5£ 105 generations to allow likelihood valuesto reach stationarity.

3. Results

3.1. Commonalities among analyses

Certain relationships were consistently recovered inall analyses under all weighting schemes, and they arebrieXy considered Wrst. OssiWed group lamprologines,represented here by 20 species after removal of duplicate

430 R. Schelly et al. / Molecular Phylogenetics and Evolution 38 (2006) 426–438

representatives, were recovered as monophyletic in eachanalysis. In all cases, L. cunningtoni nested with othernon-ossiWed lamprologines, and was sister to N. mode-stus in all but the nuclear-only analysis. Some of theNeolamprologus species suggested for inclusion in thegenus Lepidiolamprologus by Stiassny (1997), namelyN. meeli, N. boulengeri, and N. hecqui, invariably nestedin a clade of 10 taxa that included several Lepidiolamp-rologus, hereafter referred to as the two-pore Lepidio-lamprologus clade, and was always resolved as the sistergroup to (L. callipterus (A. compressiceps, A. calvus)).However, the distinctive Lamprologus lemairii, anothercandidate for inclusion in Lepidiolamprologus (Stiassny,1997), was never recovered in the two-pore Lepidiolamp-rologus clade.

3.2. Mitochondrial dataset

Direct optimization of the equally weighted 58 termi-nal dataset resulted in 72 equally parsimonious trees(strict consensus shown in Fig. 1) of length (L) D 1585,CI D 0.41, and RID 0.71 (calculations excluding uninfor-mative characters). The implied alignment consisted of1451 characters, 441 of which were parsimony-informa-tive. Species with multiple representatives were resolvedas monophyletic in all but one case: one of the Neolamp-rologus sp. “meeli-boulengeri” was recovered in a poly-tomy with the L. attenuatus, its sister group in theseanalyses. A congruent topology was recovered afterreduction of the dataset to include only single representa-tives of each species. The strict consensus of the 139 mostparsimonious trees (L D 1488, CI D 0.40, and RID 0.64;excluding uninformative characters) obtained for that 40terminal dataset is shown with branch support values,jackknife values, and results of the sensitivity analysis(Fig. 2), and with key morphological characters illus-trated and mapped onto the topology (Fig. 3). Of the1442 characters in the implied alignment generated dur-ing this analysis, 409 were parsimony informative. TheML and Bayesian results (not shown) based on a manualalignment of the mitochondrial data for 40 terminalsdiVered slightly from the parsimony result in the resolu-tion of the basal lamprologine taxa Neolamprologusmoorii and N. toae, but were otherwise congruent withthe results discussed herein. Surprisingly, Lepidiolamprol-ogus nkambae was consistently resolved in a stable posi-tion outside the two-pore Lepidiolamprologus clade, asthe sister group of the clade comprising the Lamprologuscallipterus–Altolamprologus clade and the two-pore Lep-idiolamprologus clade. This result is strongly discordantwith morphological and nuclear DNA evidence.

3.3. Nuclear dataset

Direct optimization of the equally weighted 26 termi-nal dataset obtained 56 equally parsimonious trees

(strict consensus shown in Fig. 4) of length (L) D 501,CI D 0.81, and RI D 0.93 (calculations excluding uninfor-mative characters). The implied alignment consisted of2746 characters, 359 of which were parsimony-informa-tive. Due to the unavailability of fresh extractions for allspecies represented in the mitochondrial dataset, only asubset of the species in the mitochondrial dataset wereincluded here. Furthermore, we were unable to success-fully amplify every locus for every taxon in this dataset(See Table 1). Due to a lack of variation within ossiWedgroup lamprologines at many of the nuclear loci initiallysurveyed, we included, in addition to introns, somehighly length variable microsatellite loci (in particular,UME002). We interpret the phylogeny derived frommicrosatellite loci cautiously, in light of the increasedpotential for homoplasy and diminished ability to iden-tify homology in fast evolving, indel rich repeated shortmotifs. Nevertheless, we believe that the large indelevents at microsatellite loci oVer a promising source ofsignal in the absence of suYcient variation elsewhere inthe nuclear genome. In this instance, our combinednuclear dataset was strongly discordant with themitochondrial dataset regarding the placement ofLepidiolamprologus nkambae. These nuclear data sup-port the placement of L. nkambae in an unresolved two-pore Lepidiolamprologus clade along with its probablesister species, L. kendalli, in agreement with morphologi-cal evidence.

3.4. Combined nuclear and mitochondrial dataset

When the nuclear and mitochondrial data were com-bined (tree not shown), neither of the conXicting signalswas completely swamped. The nuclear signal was suY-

cient to resolve Lepidiolamprologus nkambae with thetwo-pore Lepidiolamprologus clade, but the conXictingmitochondrial signal prevented L. nkambae from nestingwith L. kendalli. Rather, L. nkambae was resolved as sis-ter to the remaining two-pore Lepidiolamprologus.

3.5. Phylogenetic implications

These analyses consistently support the placement ofL. cunningtoni outside of the “ossiWed-group,” farremoved from the other Lepidiolamprologus with whichit was allied on the basis of a similar gestalt that evolvedin parallel. Further, the monophyly of Stiassny’s (1997)“ossiWed-group” is strongly supported by both thenuclear and mitochondrial datasets. Within the “ossi-Wed-group,” however, the mitochondrial and nucleardatasets manifest strikingly incongruent signal for thespecies Lepidiolamprologus nkambae, and morphology isstrongly at odds with the mitochondrial signal. Thisresult is interpreted as evidence of introgression and Wxa-tion of a distantly related mitochondrial haplotype inL. nkambae.

R. Schelly et al. / Molecular Phylogenetics and Evolution 38 (2006) 426–438 431

4. Discussion

4.1. On Lepidiolamprologus

Taken together, our analyses fully support the mono-phyly of Stiassny’s (1997) ossiWed group, in accord withthe more limited sampling of previous molecular studies(e.g., Nishida, 1997; Sturmbauer et al., 1994). These anal-

yses also conWrm that similarities in shape and squama-tion between L. cunningtoni and other members of Poll’s(1986) Lepidiolamprologus are superWcial, and placeL. cunningtoni in a non-ossiWed clade far removed fromthe two-pore Lepidiolamprologus clade, a Wnding inaccord with Stiassny (1997) and Takahashi (2001).Among the morphological characters supporting themolecular evidence are absence of a labial bone and a

Fig. 1. Strict consensus of 72 equally most parsimonious trees of L D 1585, CI D 0.41, and RI D 0.71 (calculations excluding uninformative characters)obtained for 58 terminal mitochondrial dataset analyzed using direct optimization with a gap:tv:ts ratio of 1:1:1. Branch support (Bremer) values areshown above nodes and jackknife values below.

432 R. Schelly et al. / Molecular Phylogenetics and Evolution 38 (2006) 426–438

single median frontal pore of the neurocranial lateralline foramina (NLF0) in L. cunningtoni. Like L. cunning-toni, most Neotropical and African cichlids have coa-lesced pores at NLF0, but some derived lamprologines(two-pore Lepidiolamprologus and Lamprologus lemairii)atavistically exhibit the condition found in distantlyrelated Malagasy ptychochromines, of two separatepores at NLF0 (Stiassny, 1992; see Fig. 3 for distributionin lamprologines).

In our mitochondrial dataset, our two-pore Lepidio-lamprologus clade is resolved sister to (L. callipterus(A. calvus, A. compressiceps)). All three are ossiWedgroup members with a single median frontal pore, andtheir placement with respect to Lepidiolamprologus is

plausible when considered in light of a preliminary mor-phological dataset (Schelly, submitted). As to the actualcomposition of Lepidiolamprologus, this study supportsthe inclusion of L. attenuatus, L. profundicola, L. elonga-tus, and L. kendalli, all of which share numerous derivedmorphological characters including a characteristicsupraoccipital crest proWle, widely spaced canines, andelevated lateral line scale counts. Also resolved in thetwo-pore Lepidiolamprologus clade are N. variostigmaand an undescribed Lepidiolamprologus, for whichcleared and stained material is not yet available to codeinternal anatomical characters. Finally, this study sup-ports Stiassny’s (1997) suggested placement of Neolamp-rologus meeli, N. boulengeri, and N. hecqui, as well as a

Fig. 2. Strict consensus of 139 trees of length 1488 (CI D 0.40, RI D 0.64) resulting from analysis of the 40 terminal mitochondrial dataset using directoptimization with a gap:tv:ts ratio of 1:1:1. Branch support (Bremer) values are shown above nodes and jackknife values below. Rectangular boxes ateach node depict results of sensitivity analysis of parameter variation, with a key to the parameter set represented by each cell at bottom-left. Shadedcells indicate that the clade in question was recovered as monophyletic for the relevant parameter set; unshaded cells indicate either the collapse ofthe node or variation (at the level of presence or absence) in the terminals recovered at that node.

R. Schelly et al. / Molecular Phylogenetics and Evolution 38 (2006) 426–438 433

new species (N. sp. “meeli-boulengeri”) thought to beclosely allied with N. meeli and N. boulengeri, in thegenus Lepidiolamprologus. For lack of available tissuematerial, the study is mute on the placement ofN. pleuromaculatus, also listed in Stiassny’s expandedLepidiolamprologus, and characterized by a “lepidio-lamprologine” ethmovomerine type (Takahashi, 2001).All Wve of these putative Lepidiolamprologus possess alabial bone and two distinct pores at NLF0.

Only in the case of Lepidiolamprologus nkambae andLamprologus lemairii does this study contradict Stiassny(1997). In the case of L. lemairii, a morphologicallyderived species without obvious aYnities to Lepidio-lamprologus, the placement outside of the two-pore Lep-idiolamprologus clade in this treatment seems to be duesimply to homoplasy for the NLF0 morphological char-acter. For instance, L. lemairii lacks the “lepidiolamprol-ogine” ethmovomerine type (Takahashi, 2001) and the

Fig. 3. MP topology from Fig. 2, showing taxa conWrmed or suspected of having two pores at NLF0 (taxa in bold, with asterisks), and Stiassny’s(1997) ossiWed group (gray box). Dorsal views of neurocrania show contrasting states at NLF0 (A, B). Lateral views of lower jaw show unossiWed (C)and ossiWed (D) conditions.

434 R. Schelly et al. / Molecular Phylogenetics and Evolution 38 (2006) 426–438

long, slender gill rakers characteristic of Lepidiolamprol-ogus. The nuclear dataset supports this interpretation,suggesting that separation of the pores at NLF0 evolvedin parallel in L. lemairii and the two-pore Lepidiolamp-rologus. However, parallel evolution of the atavisticNLF0 character state in Lepidiolamprologus nkambaeseems much less plausible in light of other morphologi-cal evidence and the nuclear dataset, casting doubt onthe ability of the mtDNA data to reveal the completeevolutionary picture and strongly implicating introgres-sive hybridization as the mechanism for the conXictingsignal. This possibility is considered below in detail.

Although several modiWcations to the current classiW-

cation currently seem warranted from the results pre-sented here, we will resist making formal changes to thegeneric classiWcation of lamprologines, pending a morethorough morphology-based analysis of the group withincreased taxon sampling, which is underway (Schelly, inpreparation).

4.2. Introgressive hybridization

Discordant signal between diVerent datasets, such asnuclear and mitochondrial DNA, can derive simply

Fig. 4. Strict consensus of 56 trees of length 501 (CI D 0.81, RI D 0.93)resulting from analysis of the 26 terminal nuclear dataset using directoptimization with a gap:tv:ts ratio of 1:1:1. Branch support (Bremer)values are shown above nodes and jackknife values below.

from homoplasy or from stochastic inheritance of ances-tral polymorphisms (McCracken and Sorenson, 2005).Hybridization resulting in introgression, or the incorpo-ration of alien genes from one species into another, canalso sometimes explain discordance between charactersets, including morphology, allozymes, mtDNA, andnuclear DNA (Dowling and DeMarais, 1993). Introgres-sive hybridization is well accepted as a fairly commonmechanism for diversiWcation in plants, but only recentlyhas the phenomenon begun to be recognized in animals(Arnold, 1997; Barton, 2001; Cathey et al., 1998; Dow-ling and Secor, 1997; Gerber et al., 2001; Roca et al.,2005; Sota et al., 2001; Sullivan et al., 2002, 2004).Among Wshes, hybridization in cichlids in particular hasbeen documented by numerous researchers (e.g., Rüberet al., 2001; Schliewen and Klee, 2004; Smith et al., 2003;Streelman et al., 2004). A typical example is reported byRognon and Guyomard (2003), who observed congruentsignal in nuclear DNA and morphology for two speciesof Oreochromis in West Africa, but found starkly con-Xicting signal in mtDNA, to the degree that some identi-cal mitochondrial sequences were found in both species.They attributed this conXicting signal to diVerentialintrogression. Introgressive hybridization resulting inspeciation has already been suggested for non-ossiWedlamprologines (Salzburger et al., 2002b). In that study,N. marunguensis was demonstrated to have a mosaic ofnuclear alleles, all Wxed, derived from two parental lin-eages, but was polymorphic for the mitochondrial con-trol region and the cytochrome b gene, also from bothparental species. Given the cases already documented incichlids, we consider that the results of this analysis sug-gest similar mechanisms.

The most striking anomaly in our mitochondrial treeis the placement of Lepidiolamprologus nkambae outsideof the two-pore Lepidiolamprologus clade2—far removedfrom L. kendalli, a species with which it shares a strikingmorphological similarity—as the sister group of theLamprologus callipterus–Altolamprologus clade and thetwo-pore Lepidiolamprologus clade. This surprisingresult was conWrmed by including four representatives ofboth species, several of which were recently collectedalong the Zambian coast of Lake Tanganyika in March,2004. Lepidiolamprologus nkambae and L. kendalli wereindependently described within a few months of oneanother (Poll and Stewart, 1977; Staeck, 1978), and theymight have been considered races of the same species ifthe authors had combined their material. Morphologi-cally, the two species are eVectively indistinguishable,and taken together, they are quite distinct from other

2 Lepidiolamprologus cunningtoni is resolved even further from theLepidiolamprologus clade among non-ossiWed lamprologines. Nonethe-less, morphological data (Schelly, submitted; Stiassny, 1997) supportsuch a placement, which will be reXected in a forthcoming revised clas-siWcation (Schelly, in preparation).

R. Schelly et al. / Molecular Phylogenetics and Evolution 38 (2006) 426–438 435

lamprologines. With identical osteology, overlappingvertebral counts, Wn spine and ray counts, scale counts,and morphometric measurements, and a striking mot-

tled color pattern that diVers considerably less betweenthe two species than the range of color variation seenwithin many lamprologine species, L. nkambae and

Table 1GenBank accession numbers for the sequences used in this study

Taxa Control region ND2 S7-1 S7-2 TmoM25 TmoM27 UME002

Altolamprologus calvus 1 DQ054913 DQ055011 DQ055072 DQ055097 DQ054967 DQ054990 DQ055117Altolamprologus calvus 2 DQ055073 DQ054991Altolamprologus compressiceps DQ054924 DQ055022 DQ055075 DQ054969 DQ054993Eretmodus cyanostictus DQ054911 DQ055010 DQ055102Julidochromis marlieri DQ054940 DQ055039 DQ055080 DQ054974 DQ054999 DQ055123Lamprologus callipterus DQ054925 DQ055023 DQ055082 DQ055104 DQ054976 DQ055001 DQ055125Lamprologus lemairii 1 DQ054921 DQ055019 DQ055063 DQ055088 DQ054958 DQ054981 DQ055109Lamprologus lemairii 2 DQ054953 DQ055056 DQ055078 DQ055101 DQ054972 DQ054997 DQ055121Lamprologus meleagris DQ054929 DQ055027Lamprologus speciosus DQ054934 DQ055032Lamprologus teugelsi DQ054955 DQ055059Lepidiolamprologus “meeli-boulengeri” 1 DQ054939 DQ055038Lepidiolamprologus “meeli-boulengeri” 2 DQ054951 DQ055052Lepidiolamprologus “sp.n.” DQ054944 DQ055045 DQ055068 DQ055093 DQ054963 DQ054986Lepidiolamprologus attenuatus 1 DQ054912 DQ055036Lepidiolamprologus attenuatus 2 DQ054938 DQ055037Lepidiolamprologus attenuatus 3 DQ054952 DQ055055 DQ055085 DQ055107 DQ054978 DQ055004 DQ055126Lepidiolamprologus attenuatus 4 DQ055057Lepidiolamprologus cunningtoni 1 DQ054919 DQ055017Lepidiolamprologus cunningtoni 2 DQ055053 DQ055084 DQ055106 DQ055003Lepidiolamprologus cunningtoni 3 DQ055054Lepidiolamprologus elongatus 1 DQ054923 DQ055021Lepidiolamprologus elongatus 2 DQ054948 DQ055049 DQ055074 DQ055098 DQ054968 DQ054992 DQ055118Lepidiolamprologus kendalli 1 DQ055060Lepidiolamprologus kendalli 2 DQ054941 DQ055042 DQ055064 DQ055089 DQ054959 DQ054982 DQ055110Lepidiolamprologus kendalli 3 DQ054942 DQ055043 DQ055065 DQ055090 DQ054960 DQ054983 DQ055111Lepidiolamprologus kendalli 4 DQ055044 DQ055066 DQ055091 DQ054961 DQ054984 DQ055112Lepidiolamprologus nkambae 1 DQ054937 DQ055035 DQ055087 DQ055108 DQ054980 DQ055005 DQ055127Lepidiolamprologus nkambae 2 DQ054943 DQ055067 DQ055092 DQ054962 DQ054985 DQ055113Lepidiolamprologus nkambae 3 DQ054945 DQ055046 DQ055069 DQ055094 DQ054964 DQ054987 DQ055114Lepidiolamprologus nkambae 4 DQ054946 DQ055047 DQ055070 DQ055095 DQ054965 DQ054988 DQ055115Lepidiolamprologus nkambae 5 DQ054947 DQ055048 DQ055071 DQ055096 DQ054966 DQ054989 DQ055116Lepidiolamprologus profundicola DQ054927 DQ055025 DQ055076 DQ055099 DQ054970 DQ054994 DQ055119Neolamprologus boulengeri 1 DQ054936 DQ055034Neolamprologus boulengeri 2 DQ055040Neolamprologus brevis DQ054922 DQ055020Neolamprologus brichardi DQ054917 DQ055015 DQ055081 DQ055103 DQ054975 DQ055000 DQ055124Neolamprologus buescheri DQ054935 DQ055033Neolamprologus caudopunctatus DQ054926 DQ055024Neolamprologus christyi DQ054954 DQ055058Neolamprologus cylindricus DQ054933 DQ055031Neolamprologus hecqui 1 DQ054920 DQ055018Neolamprologus hecqui 2 DQ055041Neolamprologus helianthus DQ054915 DQ055013Neolamprologus marunguensis DQ054916 DQ055014Neolamprologus meeli DQ054950 DQ055051 DQ055086 DQ054979 DQ054996Neolamprologus modestus DQ054914 DQ055012 DQ055079 DQ054973 DQ054998 DQ055122Neolamprologus moorii DQ054918 DQ055016 DQ055083 DQ055105 DQ054977 DQ055002Neolamprologus pulcher DQ054957 DQ055062Neolamprologus similis DQ054932 DQ055030Neolamprologus toae DQ054949 DQ055050Neolamprologus tretocephalus DQ054928 DQ055026 DQ055077 DQ055100 DQ054971 DQ054995 DQ055120Neolamprologus variostigma 1 DQ054930 DQ055028Neolamprologus variostigma 2 DQ054931 DQ055029Perissodus microlepis DQ054907 DQ055006Spathodus erythrodon DQ054909 DQ055008Tanganicodus irsacae DQ054908 DQ055007Telmatochromis bifrenatus DQ054910 DQ055009Telmatochromis temporalis DQ054956 DQ055061

436 R. Schelly et al. / Molecular Phylogenetics and Evolution 38 (2006) 426–438

L. kendalli are undoubtedly sister species on the basis ofmorphology. Konings (1998) suggested that the two spe-cies are in fact conspeciWc.

Our tree derived from nuclear data (Fig. 4) accordswell with morphological evidence; the single L. nkam-bae resolved sister to L. n.sp. may result from incom-plete lineage sorting or microsatellite homoplasy, thesignal of which could be disproportionate due to theextremely limited variability within the nuclear genomeof the two-pore Lepidiolamprologus clade. A resultbased on mtDNA that is so at odds with morphologyand nuclear DNA strongly suggests introgressivehybridization, commonly attributed as a cause of dis-cordance between phylogenies based on mitochondrialversus nuclear DNA (e.g., Salzburger et al., 2002b; See-hausen, 2004; Shaw, 2002). If introgression was at workin L. nkambae, a plausible scenario could be that thecommon ancestor of L. nkambae and L. kendalli, ormore likely the population in the Nkamba Bay areaonly, hybridized with another species and gained a sec-ond mitochondrial haplotype. Subsequently a hybridspecies retaining the morphology of L. nkambae/L. ken-dalli emerged, and the mitochondrial haplotype under-went Wxation in L. nkambae, while the introgressednuclear genes were lost due to backcrossing. Such a sce-nario seems more plausible considering that the rangeof L. nkambae encompasses a bay made turbid by theinXow of the Lufubu River in an otherwise clear lake,and turbidity has been demonstrated to interfere withspecies recognition in spawning lacustrine cichlids (See-hausen et al., 1997). This scenario would require a rela-tively long period of time passing since the hybridizationevent to allow for complete lineage sorting. The lack of aclosely related sister species of L. nkambae might beexplained by sampling bias, or by the extinction of theparental species. The alternative hypothesis, that theL. nkambae haplotype traces to diVerential Wxation of anancestral mitochondrial-haplotype polymorphism inL. nkambae vs. all species of the clade sister to L. nkambae(the Lamprologus callipterus–Altolamprologus clade andthe two-pore Lepidiolamprologus clade), seems less likely.

Acknowledgments

For support and assistance, we thank (at theAMNH) Melanie Stiassny, Scott Schaefer, John Sparks,Kevin Tang, and Leo Smith. For help with species iden-tiWcations, we thank Jos Snoeks (MRAC, Tervuren,Belgium). For providing tissues or for assistance in theWeld, we thank Heinz Büscher, Roger Bills, Alex Chilala,Cyprian Kapasa, Harris Phiri, and the team at the Mpu-lungu Station of the Department of Fisheries, Ministryof Agriculture, and Cooperatives, Republic of Zambia.R.S. was supported by the AMNH Axelrod Fund and agrant from the American Cichlid Association. W.S. was

supported by the European Union (Marie Curie Fel-lowship), the Landesstiftung Baden–Wuerttemberg andthe Deutsche Forschungsgemeinschaft. S.K., N.D., andC.S. were supported by the Austrian Science Founda-tion (Grant 15239). W.S., S.K., and N.D. got furthersupport from the Austrian Academy of Sciences [DOCand DOC-FFORTE (women in research and technol-ogy) fellowships] and the University of Graz (S.K. andN.D.).

References

Arnold, M.L., 1997. Natural Hybridization and Evolution. OxfordUniversity Press, Oxford, New York.

Barton, N.H., 2001. The role of hybridization in evolution. Mol. Ecol.10, 551–568.

Boulenger, G.A., 1898. Report on the Wshes recently obtained by Mr.J.E.S. Moore in Lake Tanganyika. Proc. Zool. Soc. Lond. 1898(pt.3), 494–497.

Boulenger, G.A., 1915. Catalogue of the Freshwater Fishes of Africa inthe British Museum (Natural History). Vol. III. British Museum(Natural History), London.

Bremer, K., 1988. The limits of amino acid sequence data in angio-sperm phylogenetic reconstruction. Evolution 42, 795–803.

Bremer, K., 1995. Branch support and tree stability. Cladistics 10, 295–304.

Cathey, J.C., Bickham, J.W., Patton, J.C., 1998. Introgressive hybridiza-tion and nonconcordant evolutionary history of maternal andpaternal lineages in North American deer. Evolution 52, 1224–1229.

Chow, S., Hazama, K., 1998. Universal PCR primers for S7 ribosomalprotein gene introns in Wsh. Mol. Ecol. 7, 1247–1263.

Colombe, J., Allgayer, R., 1985. Description de Variabilichromis, Neo-lamprologus, et Paleolamprologus genres nouveaux du Lac Tangan-ika, avec redescription des genres Lamprologus Schilthuis, 1891 etLepidiolamprologus Pellegrin, 1904 (Pisces: Teleostei: Cichlidae).Rev. Fr. Cichlidophile 49, 9–16 21–28.

De Vos, L., Seegers, L., Taverne, L., Thys van den Audenaerde, D.,2001. L’ichtyofaune du bassin de la Malagarasi (Système du LacTanganyika): une synthèse de la connaissance actuelle. Ann. Mus.R. Afr. Centr. Sci. Zool. 285, 117–135.

Dowling, T.E., DeMarais, B.D., 1993. Evolutionary signiWcance ofintrogressive hybridization in cyprinid Wshes. Nature 362, 444–446.

Dowling, T.E., Secor, C.L., 1997. The role of hybridization and intro-gression in the diversiWcation of animals. Annu. Rev. Ecol. Syst. 28,593–619.

Farris, J.S., 1989. The retention index and the rescaled consistencyindex. Cladistics 5, 417–419.

Farris, J.S., Albert, V.A., Källersjö, M., Lipscomb, D., Kluge, A.G.,1996. Parsimony jackkniWng outperforms neighbor-joining. Cladis-tics 12, 99–124.

Gerber, A.S., Tibbets, C.A., Dowling, T.E., 2001. The role of introgres-sive hybridization in the evolution of the Gila robusta complex(Teleostei: Cyprinidae). Evolution 55, 2028–2039.

Giannini, N.P., Simmons, N.B., 2003. A phylogeny of megachiropteranbats (Mammalia: Chiroptera: Pteropodidae) based on direct opti-mization analysis of one nuclear and four mitochondrial genes.Cladistics 19, 496–511.

Giribet, G., 2001. Exploring the behavior of POY, a program for directoptimization of molecular data. Cladistics 17, S60–S70.

Giribet, G., Ribera, C., 2000. A review of arthropod phylogeny: newdata based on ribosomal DNA sequences and direct character opti-mization. Cladistics 16, 204–231.

R. Schelly et al. / Molecular Phylogenetics and Evolution 38 (2006) 426–438 437

GoloboV, P.A., 1999a. NONA (NO NAME) Ver. 3. Fundacióne Insti-tuto Miguel Lillo, Tucumán, Argentina.

GoloboV, P.A., 1999b. Analyzing large data sets in reasonable times:solutions for composite optima. Cladistics 15, 415–428.

Huelsenbeck, J.P., Ronquist, F., 2001. MrBayes: Bayesian inference ofphylogeny. Bioinformatics 17, 754–755.

Kluge, A.G., Farris, J.S., 1969. Quantitative phyletics and the evolutionof anurans. Syst. Zool. 18, 1–32.

Kocher, T.D., Thomas, W.K., Meyer, A., Edwards, S.V., Pääbo, S., Vil-lablanca, F.X., Wilson, A.C., 1989. Dynamics of mitochondrialDNA evolution in animals: ampliWcation and sequencing with con-served primers. Proc. Natl. Acad. Sci. USA 86, 6196–6200.

Kocher, T.D., Conroy, J.A., McKaye, K.R., StauVer, J.R., Lockwood,S.F., 1995. Evolution of NADH dehydrogenase subunit 2 in EastAfrican cichlid Wsh. Mol. Phylogenet. Evol. 4, 420–432.

Konings, A., 1998. Tanganyika Cichlids in Their Natural Habitat.Cichlid Press, El Paso, Texas.

McCracken, K.G., Sorenson, M.D., 2005. Is homoplasy or lineage sort-ing the source of incongruent mtDNA and nuclear gene trees in thestiV-tailed ducks (Nomonyx-Oxyura)?. Syst. Biol. 54, 35–55.

Meyer, A., Morrissey, J.M., Schartl, M., 1994. Recurrent origin of asexually selected trait in Xiphophorus Wshes inferred from a molecu-lar phylogeny. Nature 368, 539–542.

Nixon, K.C., 1999. The parsimony ratchet, a new method for rapid par-simony analysis. Cladistics 15, 407–414.

Nixon, K.C., 2002. WinClada. Published by the author, Ithaca, NY,USA

Nishida, M., 1997. Phylogenetic relationships and evolution of Tang-anyikan cichlids: A molecular perspective. In: Kawanabe, H., Hori,M., Nagoshi, M. (Eds.), Phylogenetic Relationships and Evolutionof Tanganyikan Cichlids: a Molecular Perspective. Kyoto Univer-sity Press, Kyoto, pp. 3–26.

Parker, A., KornWeld, I., 1996. Polygynandry in Pseudotropheus zebra,a cichlid Wsh from lake Malawi. Environ. Biol. Fish. 47, 345–352.

Pellegrin, J., 1904. Contribution a l’étude anatomique, biologique ettaxinomique des poissons de la famille des cichlidés. Mém. Soc.Zool Fr. 16, 41–402.

Poll, M., 1974. Contribution á la faune ichthyologique du lac Tangan-ika, d’après les récoltes de P. Brichard. Rev. Zool. Afr. 88, 99–110.

Poll, M., 1986. ClassiWcation des Cichlidae du lac Tanganika. Tribusgenres et espèces. Acad. R. Belg. Mém. Cl. Sci. 45 (2), 1–163.

Poll, M., Stewart, D., 1977. Un nouveau Lamprologus du sud du LacTanganika (Zambia) (Pisces, Cichlidae). Rev. Zool. Afr. 91 (4),1047–1056.

Posada, D., Crandall, K.A., 1998. ModelTest: Testing the model ofDNA substitution. Bioinformatics 14, 817–818.

Regan, C.T., 1920. The classiWcation of the Wshes of the family Cichli-dae.–I. The Tanganyika genera. Ann. Mag. Nat. Hist. 5 (9), 33–53.

Regan, C.T., 1922. The classiWcation of the Wshes of the family Cichli-dae.–II. On African and Syrian genera not restricted to the greatlakes. Ann. Mag. Nat. Hist. 10 (9), 249–264.

Roca, A.L., Georgiadis, N., O’Brien, S.J., 2005. Cytonuclear genomicdissociation in African elephant species. Nat. Genet. 37, 96–100.

Rognon, X., Guyomard, R., 2003. Large extent of mitochondrial DNAtransfer from Oreochromis aureus to O. niloticus in West Africa.Mol. Ecol. 12, 435–445.

Rüber, L., Meyer, A., Sturmbauer, C., Verheyen, E., 2001. Populationstructure in two sympatric species of the Lake Tanganyika cichlidtribe Eretmodini: evidence for introgression. Mol. Ecol. 10,1207–1225.

Salzburger, W., Meyer, A., Baric, S., Verheyen, E., Sturmbauer, C.,2002a. Phylogeny of the Lake Tanganyika cichlid species Xock andits relationship to the central and East African haplochromine cich-lid Wsh faunas. Syst. Biol. 51, 113–135.

Salzburger, W., Baric, S., Sturmbauer, C., 2002b. Speciation via intro-gressive hybridization in East African cichlids?. Mol. Ecol. 11,619–625.

Schilthuis, L., 1891. On a collection of Wshes from the Congo; withdescription of some new species. Tijd. Nederl. Dierk. Ver. 3, 83–91.

Schelly, R., Submitted. Lamprologine phylogenetics: insights frommorphology

Schelly, R.C., Stiassny, M.L.J., 2004. Revision of the Congo riverLamprologus Schilthuis, 1891 (Teleostei: Cichlidae), with descrip-tions of two new species. Am. Mus. 3451, 1–40.

Schelly, R., Stiassny, M.L.J., Seegers, L., 2003. Neolamprologus devosisp. n., a new riverine lamprologine cichlid (Teleostei, Cichlidae)from the lower Malagarasi River, Tanzania. Zootaxa 373, 1–11.

Schliewen, U.K., Klee, B., 2004. Reticulate sympatric speciation inCameroonian crater lake cichlids. Front. Zool., 1–5.

Seehausen, O., 2004. Hybridization and adaptive radiation. TrendsEcol. Evol. 19 (4), 198–207.

Seehausen, O., van Alphen, J.J.M., Witte, F., 1997. Cichlid Wsh diversitythreatened by eutrophication that curbs sexual selection. Science277, 1808–1811.

Shaw, K.L., 2002. ConXict between nuclear and mitochondrial DNAphylogenies of a recent species radiation: what mtDNA reveals andconceals about modes of speciation in Hawaiian crickets. Proc.Natl. Acad. Sci. USA 99 (25), 16122–16127.

Sorenson, M.D., 1996. TreeRot. University of Michigan, Ann Arbor.Sota, T., Ishikawa, R., Ujiie, M., Kusumoto, F., Vogler, A.P., 2001.

Extensive trans-species mitochondrial polymorphisms in the cara-bid beetles Carabus subgenus Ohomopterus caused by repeatedintrogressive hybridization. Mol. Ecol. 10, 2833–2847.

Smith, P.F., Konings, A., KornWeld, I., 2003. Hybrid origin of a cichlidpopulation in Lake Malawi: implications for genetic variation andspecies diversity. Mol. Ecol. 12, 2497–2504.

Staeck, W., 1978. Ein neuer cichlide aus dem südlichen Tanganjikasee:Lamprologus nkambae n. sp. (Pisces, Cichlidae). Rev. Zool. Afr. 92(2), 436–441.

Stiassny, M.L.J., 1992. Atavisms, phylogenetic character reversals, andthe origin of evolutionary novelties. Neth. J. Zool. 42, 260–276.

Stiassny, M.L.J., 1997. A phylogenetic overview of the lamprologinecichlids of Africa (Teleostei, Cichlidae): a morphological perspec-tive. S. Afr. J. Sci. 93, 513–523.

Streelman, J.T., Gmyrek, S.L., Kidd, M.R., Kidd, C., Robinson, R.L.,Hert, E., Ambali, A.J., Kocher, T.D., 2004. Hybridization and con-temporary evolution in an introduced cichlid Wsh from lake Malawinational park. Mol. Ecol. 13, 2471–2479.

Sturmbauer, C., Verheyen, E., Meyer, A., 1994. Mitochondrial phylog-eny of the Lamprologini, the major substrate spawning lineage ofcichild [sic] Wshes from Lake Tanganyika in Eastern Africa. Mol.Biol. Evol. 11 (4), 691–703.

Sullivan, J.P., Lavoué, S., Hopkins, C.D., 2002. Discovery and phyloge-netic analysis of a riverine species Xock of African electric Wshes(Mormyridae: Teleostei). Evolution 56, 597–616.

Sullivan, J.P., Lavoué, S., Arnegard, M.E., Hopkins, C.D., 2004. AFLPsresolve phylogeny and reveal mitochondrial introgression within aspecies Xock of African electric Wsh (Mormyroidea: Teleostei). Evo-lution 58, 825–841.

SwoVord, D.L., 2002. PAUP*. Phylogenetic Analysis Using Parsimony(* and Other Methods). Version 4.0b10. Sinauer Associates, Sun-derland, Massachusetts

Takahashi, T., 2001. Description of the ethmovomerine region inTanganyikan cichlid Wshes (Teleostei: Perciformes). Bull. Fish. Sci.Hokkaido Univ. 52, 117–124.

Takahashi, T., 2003. Systematics of Tanganyikan cichlid Wshes (Teleo-stei: Perciformes). Ichthyol. Res. 2003, 367–382.

Takahashi, K., Terai, Y., Nishida, M., Okada, N., 1998. A novel familyof short interspersed repetitive elements (SINEs) from cichlids: thepatterns of insertion of SINEs at orthologous loci support the pro-posed monophyly of four major groups of cichlid Wshes in LakeTanganyika. Mol. Biol. Evol. 15, 391–407.

Thompson, J.D., Higgins, D.G., Gibson, T.J., 1994. CLUSTAL W:improving the sensitivity of progressive multiple sequence

438 R. Schelly et al. / Molecular Phylogenetics and Evolution 38 (2006) 426–438

alignment through sequence weighting, position-speciWc gap penal-ties, and weight matrix choice. Nucleic Acids Res. 22, 4673–4680.

Walsh, P.S., Metzger, D.A., Higuchi, R., 1991. Chelex 100 as a mediumfor simple extraction of DNA for PCR-based typing from forensicmaterial. Biotechniques 10, 506–513.

Wheeler, W.C., 1995. Sequence alignment, parameter sensitivity, andthe phylogenetic analysis of molecular data. Syst. Biol. 44, 321–331.

Wheeler, W., 1996. Optimization alignment: the end of multiplesequence alignment in phylogenetics?. Cladistics 12, 1–9.

Wheeler, W.C., 2003a. Iterative pass optimization of sequence data.Cladistics 19, 254–260.

Wheeler, W.C., 2003b. Implied alignment: a synapomorphy-based mul-tiple-sequence alignment method and its use in cladogram search.Cladistics 19, 261–268.

Wheeler, W.C., Gladstein, D., De Laet, J., 2002. POY: PhylogenyReconstruction via Optimization of DNA Data, version 3.0. Amer-ican Museum of Natural History, New York

Zardoya, R., Vollmer, D.M., Craddock, C., Streelman, J.T., Karl, S.,Meyer, A., 1996. Evolutionary conservation of microsatellite Xank-ing regions and their use in resolving the phylogeny of cichlid Wshes(Pisces: Perciformes). Proc. R. Soc. Lond. Ser. B Biol. Sci. 263,1589–1598.