AnimalBiology ,Vol.54, No. 1, pp. 77-90 (2004) Ó KoninklijkeBrill NV ,Leiden,2004. Alsoavailable online - www.brill.nl

Functional signiŽcance of variation introphic morphology within feeding microhabitat-differentiated cichlidspecies inLake Malawi

DAUD KASSAM 1; ,DEANC.ADAMS 2 andKOSAKU Y AMAOKA 1

1 Departmentof Aquaculture,Kochi University, B 200Monobe, Nankoku-shi, Kochi, 783-8502,Japan 2 Departmentof Ecology,Evolution and Organismal Biology, Iowa State University, Ames, Iowa50010, USA

Abstract—Shapevariation in trophicmorphology between species in two trophic guilds (zooplankton andepilithic algal feeders) was investigated using landmark-based geometric morphometrics. Three disarticulatedbone elements from the head region were examined; the neurocranium, the premaxilla andlower jaw. From separate analyses of each bone element, signiŽ cant shape variation was identiŽ ed betweenspecies in each trophic guild. The deformation grids generated revealed that, for the zooplanktonfeeders, Ctenopharynxpictus hasa longerneurocranium, a longerand ventrally directed vomer,a largerorbit, a shorterascending arm, a shortermaxillad spine, and a morecompressed articularbone relative to Copadichromisborleyi .Inalgal feeders, Labeotropheusfuelleborni has ashorterneurocranium, a smallerorbit, a ventrallydirected vomer, a longerascending arm, a shorterdentigerous arm, increased height of the articular process, and a moreelongated dentary than Petrotilapiagenalutea .Observedanatomical differences are discussed in terms of function, speciŽcally with respect to thefeeding microhabitat differentiation between species in eachtrophic guild.These differences enable us to appreciate the role that trophic morphology plays in enhancing ecologicalsegregation, leading to coexistence of thespecies.

Keywords:algalfeeders; Cichlidae; geometric morphometrics; thin-plate spline; zooplankton feed- ers

INTRODUCTION Inthe East AfricanGreat Lakes,viz. Victoria, Tanganyikaand Malawi, many cichlid species are knownto coexist inhigh densities alongthe rockyshores. Such coexistence is frequentlyattributed tothe mannerin which these cichlids partition

Correspondingauthor; e-mail: [email protected] 78 D.Kassam,D.C. Adams& K.Yamaoka resources throughtemporal, spatial andtrophic means (Ribbinket al., 1983;Witte, 1984;Bouton et al., 1997).Trophically, cichlids are knownto segregate along variousniche axes including:food size partitioning,quantitative differences in food composition,differences in foodcollecting strategies, andpartitioning of feeding microhabitats (Yamaoka,1982, 1997; Hori, 1983, 1991; Witte 1984;Goldschmidt, 1990;Reinthal, 1990;Y uma,1994; Kohda and T anida,1996; Genner et al., 1999a, b).In most cases, these trophically segregatedgroups can be identiŽ ed by structural differences in their trophicmorphology ,eventhough such differentiation is related moreto the waythe foodis capturedand processed than to the typeof food consumed(Barel, 1983;Y amaoka,1997). The keyto trophicsegregation in cichlids appearsto bethe diversiŽcation ofthe oral jaw apparatusthat has enabledcichlids to evolvespecialised modesof feeding,and to utilise almost all available feeding niches. InLake Malawi, manycichlid species coexist alongthe rockyshores (e.g.,15 species at West Thumbiisland, ourcollection site; Ribbinket al., 1983).T oun- derstandfully whatmechanisms promotethe coexistence ofthese species, the role ofmorphologicalvariation within andbetween species must beinvestigated. Some studies havebegun to investigate this (e.g.Reinthal, 1989;Kassam et al., 2002a, b).Kassam et al. (2003a)examined the role ofbody shape in resource partitioning amongfour species coexisting alongLake Malawi’ s rockyshores: Copadichromis borleyi and Ctenopharynxpictus (zooplanktonfeeders), and Labeotropheusfuelle- borni and Petrotilapiagenalutea (epilithic algal feeders).These species are segre- gatedalong a foodaxis, andare segregatedspatially in terms offeedingmicrohab- itat. Ctenopharynxpictus is mainly benthophagous,but also feeds fromthe water columnwhen zooplankton is inabundance (T. Sato, pers. comm.), while its coun- terpart, Copadichromisborleyi, is reportedto feedfrom the openwater (Ribbinket al., 1983;Konings, 1990). The two algal feeders inhabit shallow rockyareas, but L.fuelleborni is commonlyfound on the sediment-free wave-beatensides ofthe rocks(Ribbink et al., 1983;Konings, 1990). Kassam et al. (2003a)found that the headregion was most morphologicallydivergent among these species. This Žnding promptedus to investigate furtherwhat speciŽ c anatomical features maybe respon- sible forthe observedvariation in headshape. W etherefore,analysed several bone elements in the headregion: the neurocranium,the lowerjaw andthe premaxilla in the upperjaw. The neurocranium was includedbecause of the role it plays in conjunctionwith the oral jaws throughthe ethmovomerregion (Reinthal, 1989;Y a- maoka,1997; Albertson and Kocher, 2001). Thegoal of our study was toexamine patterns ofshape variation inthese anatomical elements betweenspecies in eachtrophic guild, and determine whether the observedmorphological patterns relate toresource partitioning (especially feedingmicrohabitat differentiation), inan attempt to understandthe role that trophicmorphology plays as amechanism promotingspecies coexistence. Trophicmorphology of Malawian cichlids 79

MATERIALS AND METHODS Specimenpreparation Specimens (n 20per species) usedin this studywere collected fromW est ThumbiIsland as describedin Kassam et al. (2003a).The following species were used: Copadichromis borleyi (standardlength, SL, 80.5-127.5mm), Ctenopharynx pictus (SL,78.6-101.3), L.fuelleborni (SL,81.7-104.2) and P.genalutea (SL, 85.9- 116.4).W eusedPotthoff ’s(1984)protocol to clear andstain all bonesin the headregion. Drawings of all anatomical structures weremade using a Leica-MS 5 microscopeattached to acamera lucida.These were later scannedand digitised for geometric morphometricanalysis. Osteological nomenclaturefollows that ofBarel et al. (1976).

Geometric morphometricsand statistical analyses FollowingAlbertson and Kocher (2001), we focused on individual skeletal ele- ments, whichenables us toreveal patterns ofmorphologicalvariation that are other- wise obscuredif articulated skeletons orexternal morphologyalone is considered. QuantiŽcation ofthe shapeof eachdisarticulated bonestructure (neurocranium,pre- maxilla andlower jaw) was doneusing landmark-based geometric morphometrics (GM)methods (Rohlf and Marcus, 1993). First, TPSDIG32(Rohlf, 2001) was used to digitise the locations ofbiologically homologous landmarks on the lateral side ofeachbone structure (Žg. 1). For the neurocranium,the followinglandmarks were recorded(Ž g. 1a): 1)rostral tip ofthe vomer;2) caudal-most pointof the preorbital ridge; 3)dorsal tip ofthe suppraoccipital crest; 4)ventral process ofthe vertebrad concavity;5) pharyngobranchiad apophysis; 6) tip ofpostorbital process; 7)tip of preorbital process; 8)caudal-most pointof the vomerine-palatinadarticulation facet. Forthe premaxilla, Žvelandmarks were recorded (Ž g. 1b):1) dorsal process ofthe ascendingspine; 2)rostro-most pointof the dentigerousarm; 3)caudalprocess of the dentigerousarm; 4)dorsal process ofthe maxillad spine; 5)ventro-mostpoint ofthe interprocess edge.Finally, eight landmarkswere recorded from the lowerjaw (Žg. 1c): 1)rostral tip ofthe dentary;2) dorsal tip ofthe coronoid(dentary) process; 3)dorsal tip ofthe primordial (articular) process; 4)dorsal process ofthe suspenso- riad articulation facet; 5)postarticulation process (ofthe suspensoriadarticulation facet); 6)retroarticular process; 7)rostral process ofthe coulter area; 8)tip ofthe rostral process ofthe articular. Theselandmarks were chosen for their capacity to represent prominentfeatures andto capture the overall shapeand structure ofeach bone. Forall subsequentanalyses, the landmarkcoordinates from each bone were treated separately. First, all specimens weresuperimposed using the Generalised Procrustes Analysis (GPA)(Rohlfand Slice, 1990)to remove non-shape variation arising fromdifferences in scale, orientation andtranslation. Fromthe GPAaligned specimens, shapevariables wereobtained by generating partial warpscores and uniformcomponents using the thin-plate spline andstandard uniform equations 80 D.Kassam,D.C. Adams& K.Yamaoka

Figure 1. Landmarkscollected on each skeletal element; a) neurocranium, b) premaxilla, c) lower jaw.The landmarks are deŽ ned in Materialand Methods section.

(Bookstein,1989, 1991, 1996). T ogether,the uniformand non-uniform components are treated as aset ofshapevariables forstatistical comparisonsof shape variation within andamong groups (see e.g.,Caldecutt andAdams, 1998; Adams and Rohlf, 2000;Rü ber and Adams, 2001; Kassam et al., 2003a,b). In addition to this analysis, the specimens foreach species weresuperimposed separately usingGP A,and the average(consensus) conŽ guration of landmarks was obtained.These mean specimens werethen compared to the overall consensusconŽ guration (reference) to visualise shapevariation amongspecies usingthin-plate spline deformationgrids. TPSRELWsoftware (Rohlf,2002) was usedto performall these analyses. Toidentify anyshape variation amongspecies, aCanonicalV ariate Analysis (CVA)was performedon the weightmatrix ofshapevariables (partial warpscores anduniform components of shape).This analysis was performedseparately onthe shapedata foreach bone. When the multivariate analysis ofvariance (MANOV A) identiŽed signiŽ cant differences amongspecies, pairwise multiple comparisons usinggeneralised Mahalanobisdistance ( D2)wereperformed to determine which species differedfrom one another (after Bonferroniadjustment). TheCV Aanalyses wereperformed in NTSYS-PC(Rohlf, version 2.1, 2000). Trophicmorphology of Malawian cichlids 81

Table 1. Pairwisecomparisons for the three bone elements, based on Mahalanobis distances: Upper value representsneurocranium, middle value is forpremaxilla, and lower value is forlower jaw .nsdepicts pairsnot signiŽ cantly different (after Bonferroni correction).

Species C. borleyi C. pictus L.fuelleborni P.genalutea C. borleyi 0.0 C. pictus 6.2 0.0 4.0 7.1 L.fuelleborni 8.6 8.9 0.0 9.1 11.8 13.1 12.9 P.genalutea 4.5 ns 5.3 7.9 0.0 9.2 10.5 6.9 11.1 11.8 10.9

RESULTS Forall disarticulated bonestructures, MANOVArevealedsigniŽ cant differences amongspecies (neurocranium:Wilks’ ¸ 0:003997, F 29:35, P < 0:0001; premaxilla: Wilks’ ¸ 0:002955, F 76:48, P < 0:0001;lower jaw: Wilks’ ¸ 0:000289, F 79:03, P < 0:0001),and pairwise comparisonsfor each of the bonestructures revealedsigniŽ cant differences betweenspecies within eachtrophic guild(T able 1). Discrimination amongspecies canalso beinterpreted byexamining the patterns foundin an ordination of specimens in shapespace, using CV1 and CV2 as the ordinationaxes (Žg. 2a, b, c). For all anatomical structures, the Žrst CVaxis clearly discriminates betweenalgal feeders fromzooplankton feeders (i.e. between trophicguilds) while the secondCV axis discriminates species withintrophic guilds,according to feedingmicrohabitat. Theone exception to this pattern is foundfor the neurocraniumwhere P.genalutea (algal feeder)partially overlapswith Copadichromisborleyi (planktonfeeder), re ecting their similarity identiŽed in the pairwise comparisons(Table 1). Toanatomically describe the shapevariation revealedbetween species, we generatedand examined thin-plate spline deformationgrids forthe averagebone foreach species. Amongzooplankton feeders, Ctenopharynxpictus has a longer neurocranium,a longerand ventrally directed vomer,and a larger orbit,relative to Copadichromisborleyi (Žg. 3a, b). Among algal feeders, L.fuelleborni has a shorter neurocranium,and a smaller orbit andventrally directed vomer,relative to P. genautea (Žg. 3c, d). The premaxilla in zooplanktonfeeders is characterised byan ascendingarm, and a longerascending spine andmaxillad spine in Copadichromis borleyi relative to Ctenopharynxpictus (Žg. 4a, b). In contrast, in algal feeders the ascendingarm andascending spine are longer,and the dentigerousarm shorter in L.fuelleborni relative to P.genalutea (Žg. 4c, d). In general, zooplankton 82 D.Kassam,D.C. Adams& K.Yamaoka

Figure 2. Ordinationof specimens on the Ž rsttwo canonical variate axes; a) neurocranium, b)premaxilla, c) lower jaw. In all Ž gures,Ž lledcircle ( l)represents Copadichromisborleyi , Ž lled downtriangle ( ) is Ctenopharynxpictus ,opendiamond ( e ) is Labeotropheusfuelleborni , and open uptriangle( ¢) is Petrotilapiagenalutea . feeders possess longerascending arms, shorter dentigerousarms, andshorter maxillad spines thanalgal feeders.Additionally, the anglebetween ascending and dentigerousarms is acute in zooplanktonfeeders, while obtusein algal feeders (Žg. 4).Comparing lower jaws betweenzooplankton feeders showsthat the articular bone of Ctenopharynxpictus is morecompressed than that of Copadichromis borleyi (Žg. 5a,b), while foralgal feeders,in L.fuelleborni ,the articular process increases in relative height,there is amoreelongated dentary ,anda moreconcave suspensoriadarticulation facet relative tothat in P.genalutea (Žg. 5c,d). Between Trophicmorphology of Malawian cichlids 83 s i m o r h c i d a p o C ) a ; s r e d e e f . n m o u t i k n n a r a l c p o r o u o e z n o ) e w t , r a o e t f u l s n a n o i e t g a r a u i g p Ž a l n i t o o c r t s e u P s n ) e d s , i n o n r c o n b e e l e l e w t u f e b s u n e o h i t p a o i r r t a o v e b e p a a L h ) s c ; m s r u i e n d a e r e c f o l r a u g e l a n o g n w i t t a d c n i a d , n s i u s t d c i i r p g x n n o y i r t a a h m p r o o n f e e t D C ) . b 3 , i e y r e u l r g i o F b 84 D.Kassam,D.C. Adams& K.Yamaoka trophicguilds, the lowerjaw tends to bemoreelongated and slender in zooplankton feeders thanin algal feeders.

DISCUSSION Usinglandmark-based geometric morphometrictechniques, we quantiŽed the shape ofseveral anatomically important bones,and compared patterns ofvariation within andbetween trophic guilds of cichlid Žshes. Werevealedtrophic morphological variation betweenspecies ineach trophic guild (zooplankton and algal feeders). Amongzooplankton feeders, it was foundthat Ctenopharynxpictus has a longer neurocranium,a larger orbit,a shorter ascendingarm andmaxillad spine,and a morecompressed articular bonerelative to Copadichromis borleyi .Amongalgal feeders,it was revealedthat L.fuelleborni has ashorter neurocranium,a ventrally directed vomer,a longerascending arm andspine, a shorter dentigerousarm, andan increased heightof the articular process relative to P.genalutea . Some generalfeatures distinguishingthe trophicguilds were also found;for instance, zooplanktonfeeders havelonger ascending arms, shorter dentigerousarms, anacute anglebetween ascending and dentigerous arms, shorter maxillad spines, andmore elongatedand slender lowerjaws relative to algal feeders.Because these structures are related to feeding,some ofthe observeddifferences in trophicmorphology betweenspecies might reect functionaldifferences. Interms offunction, the role ofthe observeddifferences introphic morphology betweenzooplankton feeders canbe consideredas follows: the longerneurocranium in Ctenopharynxpictus implies that the volumeof its buccalcavity is greater, resulting in anattenuated andtruncated cone capable of generating the highnegative pressure requiredfor suction feeding(Liem, 1991). Its larger orbit size is directly related tolarger eyes,which concurs with the results fromlinear measurements ofeye diameter forthis species (Kassam et al., 2003a).All zooplanktonfeeders requirelarge eyes (Fryerand Iles, 1972;Hart andGill, 1994),since theyhave to visually select their prey(which tend to besmall). However,Kassam et al. (2003a)hypothesised that the larger eyes in Ctenopharynxpictus enablethe Žsh to obtainprey in abenthicenvironment, since this probablyrequires higherresolving powerthan obtaining prey in otherhabitats (as in Copadichromisborleyi ). For Copadichromisborleyi ,the longerascending arm ofthe premaxilla suggests that this species canprotrude its mouthto agreater extent thancan Ctenopharynxpictus . Sucha protrudedmouth forms atube-like structure that enables this species to suckindividual prey from the water columninto its mouth(Fryer and Iles, 1972; Greenwood,1974). Thefeeding mode and associated trophicmorphology of Copadichromis borleyi is similar to that ofa LakeT anganyikanzooplankton feeder, Gnathochromis per- maxillaris (see Fryerand Iles, 1972).Obtaining prey individually through suction feedingmay not exist in Ctenopharynxpictus becauseit sucks inloose sediments onthe bottom(from which prey is sieved),almost like avacuumcleaner (Ribbinket Trophicmorphology of Malawian cichlids 85 , i y e l r o b s i m o r h c i d a p o C ) a ; s r e d e e f n o t . k a l n l a i l x p a o o m z e r o p w ) t e r , o a f e t s u n l o a i t n a e r g u g a i Ž p n a o l i c t s o r u t s e n P e s ) n d o , i c n n r e o e b e w l t l e e b u f n s o i u t e a i h r p a o v r t e o p e a b h a s L a l ) l c i x ; s a r e m e d r e p e f g l n a i t g l a a c i o d n w i t s d d i n r a g , s n u o t i c t i a p m x r n o f y r e a D h . p 4 o n e e r t u C g i ) F b 86 D.Kassam,D.C. Adams& K.Yamaoka , i y e l r o b s i m o r h c i d a p o C ) a ; s r e d e e f n o t k . n w a a l j p r o e o z w o o l ) w t e r , o a f e t s u n l o a i t n a e r g u a g i Ž p n a l o i c t o s r u t s e n P e ) s d n , o i c n r n o e b e e l w t l e e b u f n s o u i t e a h i r p a o r v t e o e p a b h a s L ) w c a j ; s r r e e w d o e l e f g l n a i t g a l c a i d o n w i t s d d i n r a g , s n u o t i t c i a p m x r n o f y e r a D h . p 5 o n e e r t u C g i ) F b Trophicmorphology of Malawian cichlids 87 al., 1983).Unlike its counterpart, Ctenopharynxpictus doesnot form a longsuck- ingtube, but can suck in large amountsof loose sediment becauseof its larger gape (Kassam et al., 2003a).Synthesising these Žndings,it is apparentthat feedingmi- crohabitat segregation,coupled with trophicmorphological differentiation between these twozooplankton feeders, enhances ecological separation andpossibly leads to their stable coexistence. Forthe epilithic algal feeders,some patterns oftrophic diversiŽ cation may represent functionalshifts in morphology.Yamaoka(1997) classiŽ ed the Lake Tanganyikanepilithic algal feeders intotwo major trophicgroups; browsers and grazers,of which the latter has three subgroups:combers, scrappers andsuckers. Combersare Žshes that combunicellular algae,mainly diatoms, fromŽ lamentous algae (Yamaoka,1997), whereas scrappers are Žshes that scrape offepilithic algae (Yamaoka,1987), and suckers refer togroup of Ž shes that take mainlyloose aufwuchstogether with organicdebris byusing bucco-opercular cavity as asuction pump(Y amaoka,1991). Y amaoka(1997) identiŽ ed Petrochromisfasciolatus as a comberspecies and,because this is the ecomorphologicalequivalent to Petrotilapia genalutea (Kassam et al., 2003b),we concludethat P.genalutea is also acomber. Yamaoka(1997) also foundthat the neurocranium,and especially the vomer region,of the browser Tropheusmoorii was verysimilar tothat of Labeotropheus fuelleborni (Žg. 2-15, Y amaoka,1997). Additionally he found that members of these generahave ventrally directed mouths,and a verystraight jaw margin, whichextends transversely across the full widthof the head(Fryer and Iles, 1972;Y amaoka,1987). The ventrally directed mouthin L.fuelleborni was also identiŽed by Kassam et al. (2003a),and in T. moorii byY amaokaet al. (unpubl.). This resemblance introphic morphology, coupled with the similar feedinghabit displayedby the twospecies, allows usto infer that L.fuelleborni is also abrowser. Because L.fuelleborni is abrowserand P.genalutea is acomber,the two species havea markeddifferentiation in feedinghabit: acrucial factor in enhancing their ecological separation.This is consistent with earlier workshowing that understandingthe mechanisms ofresourcepartitioning in Žshes is best donethrough identifyinghow and where a Žsh feeds,rather thandetermining what food items are actually consumed(Smith andT yler,1973; Barel, 1983;Y amaoka,1997). Theventrally directed andwell-developed vomer in L.fuelleborni is therefore suggestedto have a similar functionto that in T. moorii,namely,toprovide abase forthe powerfuljaw bitingmechanism (Yamaoka,1997). The fact that L.fuelleborni feeds while its bodyis alignedhorizontally/ parallel to the substrate (Ribbinket al., 1983;Albertson and Kocher, 2001) is also in accordwith this Žnding.The horizontally directed vomerin P.genalutea mayre ect the fact that this species rostrally protrudesits mouthand combs Ž lamentous algae while the bodyis perpendicularto the substrate. Consideringthe lowerjaw, the most pronouncedvariation between L.fuelleborni and P.genalutea is onthe articular process whichis higherin the formerthan the latter. Interms offunction, Otten (1983)found that the articular process acts as 88 D.Kassam,D.C. Adams& K.Yamaoka avital lever inthe jaw adductionbecause the secondadductor mandibulae inserts onit. Thus,the higherarticular process in L.fuelleborni implies that this species cangenerate a greater force,resulting ina strongerbite (Albertsonand Kocher, 2001;Albertson et al., 2003),whereas the shorter articular process in P.genalutea implies that its jaw canclose faster duringthe combingprocess. Fast jaw closing is consistent with whatY amaoka(1983) found in P.fasciolatus ,anecomporphological equivalentof P.genalutea ,whichshowed the highest grazingspeed among its congenersin LakeTanganyika. Oneof the generalisations maderegarding differences in trophicmorphology be- tweenzooplankton and algal feeders involvesthe angleformed between the as- cendingand dentigerous arms ofthe premaxilla. This angleis acute in zooplankton feeders andobtuse in algal feeders.Otten (1983)and Bouton et al. (1999)reported that the obtuseangle, as foundin algal feeders,enables the Žsh to exert agreater forcewhen biting. Inthis studywe foundsigniŽ cant differences in the shapes ofvarious bony ele- ments ofthe trophicapparatus, both within andbetween trophic guilds. Examining eachbone separately revealedshape differences in the individualanatomical ele- ments ofthe trophicapparatus, which are obscuredwhen the trophicapparatus is examinedin anarticulated fashion.Therefore, examining these elements individ- ually canbe a useful wayof revealing potentially important functionalanatomi- cal differences amongspecies. TheseŽ ndingscompliment those ofKassam et al. (2003a),who found body shape variation amongthese species. Nevertheless, when trophicecology is considered,it is likely that bodyshape alone cannot identify all ofthe potentially meaningfulmorphological variation within andamong species. Therefore,an integrative approachthat combinesbody shape and trophic morpho- logical data is likely toprovide the best explanationof morphological diversity in these cichlids. Sucha pluralistic approachwill enableus to appreciate the role that morphologyplays in resourceuse andpartitioning, so that wecanbetter understand howthese species coexist in nature.

Acknowledgements Wewould like to acknowledgethe Malawi governmentthrough its Fisheries Departmentfor permitting us tocollect samples fromLake Malawi. DrAmbali andhis staff deserveour heartfelt thanksfor helping us incollecting samples. This research was sponsoredin part byNational Science Foundationgrant DEB-0122281 (to DCA).

REFERENCES

Adams,D.C. &Rohlf,F .J.(2000)Ecological character displacement in Plethodon:Biomechanical differencesfound from a geometricmorphometric study. Proc.Natl. Acad. Sci . USA, 97, 4106- 4111. Trophicmorphology of Malawian cichlids 89

Albertson,R.C. &Kocher,T .D.(2001) Assessing morphological differences in adaptive trait: a landmark-basedmorphometric approach. J. Exp. Zool.,289(6),385-403. Albertson,R.C., Streelman, J.T .&Kocher,T .D.(2003) Directional selection has shaped the oral jaws ofLake Malawi cichlid Ž shes. Proc.Natl. Acad. Sci . USA,100,5252-5257. Barel,C.D.N., Witte, F .&vanOijen, M.J.P .(1976)The shape of theskeletal elements in thehead of a generalizedHaplochromis species: H . elegans Treawavas1933 (Pisces, Cichlidae). Neth.J. Zool ., 26,163-265. Barel,C.D.N. (1983)T owardsa constructionalmorphology of cichlidŽ shes(Teleostei, Perciformes). Neth.J. Zool. ,33(4),357-424. Bookstein,F .L.(1989) Principal warps: thin-plate splines and the decomposition of deformations. IEEE Trans.P att.Anal. Mach. Intell. ,11,567-585. Bookstein,F .L.(1991) Morphometrictools for landmark data .CambridgeUniversity Press, New York. Bookstein,F .L.(1996b) A standardformula for the uniform shape component. In: L.F .Marcus,M. Corti,A. Loy,G. Naylor& D.Slice(Eds.), Advancesin morphometrics ,pp.153-168. Plenum Press,New York. Bouton,N., Seehausen, O. &vanAlphen, J.J.M. (1997) Resource partitioning among roc-dwelling haplochromines(Pisces: Cichlidae) from Lake Victoria. Ecol.Freshw .Fish ,6(4),219-234. Bouton,N., Witte, F .,vanAlphen, J.J.M., Schenk, A. &Seehausen,O. (1999)Local adaptations in populationsof rock-dwelling haplochromines (Pisces: Cichlidae) from southern Lake Victoria. Proc.R. Soc.Lond. B ,266,355-360. Caldecutt,W .J.&Adams,D.C. (1998)Morphometrics of trophic osteology in the threespine spine stickleback, Gasterosteusaculeatus. Copeia ,1998(4),827-838. Fryer,G. &Iles,T .D.(1972) Thecichlid Ž shesof the Great Lakes of Africa: Their Biology and Evolution. Edinburgh,Oliver and Boyd. Genner,M.J., Turner, G.F .,Barker,S. &Hawkins,S.J. (1999a) Niche segregation among Lake Malawi Žshes?Evidence from stable isotope signatures. Ecol. Lett.,2,185-190. Genner,M.J., Turner, G.F .&Hawkins,S.J. (1999b) Foraging of rockyhabitat cichlid Ž shesin Lake Malawi:coexistence through niche partitioning? Oecologia,121,283-292. Goldschimidt,P .T.,Witte,F .&deVisser, J. (1990)Ecological segregation in zooplanktivorus haplochrominespecies (Pisces: Cichlidae) from Lake Victoria. Oikos,58,343-355. Greenwood,P .H.(1974) Cichlid Ž shesof LakeVictoria, East Africa: the biology and evolution of a species ock. Bull.Br .Mus.(Nat. Hist.) Zool. Suppl. , 6, 1-134. Hart,P .J.B.& Gill,A.B. (1994)Evolution of foragingbehavior in the threespine stickleback. In: M.A. Bell& S.A. Foster(Eds.), Theevolutionary biology of the threespine stickleback ,pp.207-236. OxfordUniversity Press, Oxford. Hori,M. (1983)Feeding ecology of 13 speciesof Lamprologus(Cichlidae) coexisting at a rockyshore ofLake T anganyika. Physiol.Ecol., Jpn ,20,129-149. Hori,M. (1991)Feeding relationships among cichlid Ž shesin Lake T anganyika:effects of intra- and interspeciŽc variationsof feedingbehaviour on theircoexistence. Ecol.Int. Bull. ,19,89-101. Kassam,D.D., Sato, T .&Yamaoka,K. (2002a)Landmark-based morphometric analysis of the body shapeof two sympatric species, Ctenopharynxpictus and Otopharynx sp.“ heterodonnankhumba” (Teleostei:Cichlidae), from Lake Malawi. Ichthyol.Res .,49(4),340-345. Kassam,D.D., Sato, T .&Yamaoka,K. (2002b)Comparative morphometrics and associated growth trendsof twobenthophagous cichlid species from Lake Malawi (Pisces, Perciformes). Zool. Anz., 241(4),381-387. Kassam,D.D., Adams, D.C., Ambali, A.J.D. & Yamaoka,K. (2003a)Body shape variation in relation toresource partitioning within cichlid trophic guilds coexisting along the rocky shore of Lake Malawi. Anim. Biol., 53, 59-70. Kassam,D.D., Adams, D.C., Hori, M. &Yamaoka,K. (2003b)Morphometric analysis on ecomorpho- logicallyequivalent cichlid species from Lakes Malawi and T anganyika. J. Zool.,260,153-157. 90 D.Kassam,D.C. Adams& K.Yamaoka

Kohda,M. &Tanida,K. (1996)Overlapping territory of benthophagous cichlid Ž sh, Lobochilotes labiatus,inLake T anganyika. Environ.Biol. Fishes , 45, 13-20. Konings,A. (1990) AdKonings’ book of cichlids and all the other Ž shesof Lake Malawi . T. F. H. Publications,Neptune City, USA. Liem,K.F .(1991)Functional morphology. In: M.H.A. Keenleyside (Ed.), CichlidŽ shes:behaviour , ecologyand evolution ,pp.129-150. Chapman and Hall, London. Otten,E. (1983)The jaw mechanism during growth of a generalized Haplochromis species: H. elegans Trewavas,1933 (Pisces, Cichlidae). Neth.J. Zool. , 33, 55-98. Potthoff,T .(1984)Clearing and staining techniques. In: H.G. Moser,W .J.Richards,D.M. Cohen, M.P . Fahay,A.W .KendallJr. & S.L. Richardson(Eds.), Ontogeneticand systematic Ž shes , pp. 35-37. Specialpublication number 1. Am. Soc.Ichthyol. Herp .,AleenPress Inc., Lawrence, USA. Reinthal,P .N.(1989)The gross intestine morphology of a groupof rock-dwelling cichlid Ž shes (Pisces,Teleostei) from Lake Malawi. Neth.J. Zool. ,39,208-225. Reinthal,P .N.(1990) Morphological analyses of theneurocranium of agroupof rock-dwellingcichlid Žshes(Cichlidae, Perciformes) from Lake Malawi, Africa. Zool.J. Linn. Soc. ,98,123-139. Ribbink,A.J., Marsh, A.B., Marsh, A.C., Ribbink, A.C. &Sharp,B.J. (1983) A preliminarysurvey of thecichlid Ž shesof rockyhabitats in Lake Malawi. S.Afr.J.Zool. ,18(3),149-310. Rohlf,F .J.(2000) NTSYS-PC, NumericalT axonomySystem for the PC version2.1 .ExeterSoftware, Setauket,USA: AppliedBiostatistics Inc. Rohlf,F .J.(2001) TPSDIG32, version1.19 .Geometricmorphometric software for the PC. http://www. life.bio.sunysb.edu/morph/software.html Rohlf,F .J.(2002) TPSRELW,version1.24 .Geometricmorphometric software for the PC. http://www. life.bio.sunysb.edu/morph/software.html Rohlf,F .J.&Slice,D.E. (1990)Extensions of theProcrustes method for the optimal superimposition oflandmarks. Syst. Zool.,39,40-59. Rohlf,F .J.&Marcus,L.F .(1993)A revolutionin morphometrics. Trend.Ecol. Evol .,8,129-132. Rüber, L. &Adams,D.C. (2001)Evolutionary convergence of body shape and trophic morphology in cichlidsfrom Lake T anganyika. J.Evol.Biol .,14,325-332. Smith,C.L. &Tyler,J.C. (1973) Direct observations of resource sharing in coral reef Ž sh. Helgol. Wiss.Meeresunters ,24,264-275. Witte,F .(1984)Ecological differentiation in LakeVictoria haplochromines: Comparison of cichlid species ocksin African Lakes. In: A.A. Echelle& I.KorniŽeld (Eds.), Evolutionof Ž shspecies  ocks,pp.155-167. University of Maine at OronoPress. Yamaoka,K. (1982)Morphology and feeding behaviour of Ž vespecies of genus Petrochromis (Teleostei,Cichlidae). Physiol.Ecol. Jpn ,19(1),57-75. Yamaoka,K. (1983)Feeding behaviour and dental morphology of algae scraping cichlids (Pisces: Teleostei)in LakeT anganyika. AfricanStudy Monographs , 4, 77-89. Yamaoka,K. (1987)Comparative osteology of the jaw of algal-feedingcichlids (Pisces, Teleostei) of LakeT anganyika. Rep.Usa Mar .Biol.Inst. Kochi Univ . , 9, 87-137. Yamaoka,K. (1991)Feeding relationships. In: M.H.A. Keenleyside. (Ed.), CichlidŽ shes:behaviour , ecologyand evolution ,pp.151-172. Chapman and Hall, London. Yamaoka,K. (1997)Trophic ecomorphology of Tanganyikancichlids. In: H. Kawanabe,M. Hori& M.Nagoshi(Eds.), Fishcommunities in LakeT anganyika ,pp.25-56. Kyoto University Press. Yuma,M. (1994)Food habits and foraging behaviour of benthivorous cichlid Fishes in Lake Tanganyika. Environ.Biol. Fishes ,39,173-182.