Heredity 73(1994) 117—123 Received 25 November 1993 Genetical Society of Great Britain

Application of the RAPD technique in fish: species and subspecies identification

F. BARDAKCI* & D. 0. F. SKIBINSKI Molecular Biology Research Group, School of Biological Sciences, University of Wales, Singleton Park, Swansea SA2 8PP, U.K.

RandomAmplified Polymorphic DNA (RAPD) analysis was applied to three species of the tilapia genus and four subspecies of 0. niloticus. Thirteen random lO-mer primers were used to assay polymorphisms within and between populations. Different RAPD fragment patterns were observed for different species, although not always for different subspecies. Evidence is presented that RAPD markers might be useful for systematic investigation at the level of species and subspecies.

Keywords:DNA,Oreochromis, polymorphism, RAPD, systematics, tilapia.

fully to identify the subspecies of 0. niloticus (Capili, Introduction 1990; Seyoum & Kornfield, 1992) but little effort has Tilapiaare cultured extensively throughout the world, yet been devoted to the analysis of nuclear DNA for especially in Africa and the Far East. Most of the this purpose. important tilapiine species used in aquaculture are The development of random amplified polymorphic members of the Oreochromis genus, as many members DNA (RAPD) markers, generated by the polymerase of this taxon grow well under diverse culture condi- chain reaction (PCR), allows the examination of tions (Fryer & lIes, 1972; Pullin & Capili, 1989). genomic variation without prior knowledge of DNA Despite its commercial importance, resources have sequences (Williams et a!., 1990, 1993; Welsh & only recently been devoted to the development of McClelland, 1990; Hadrys, 1992). The number and improved strains of tilapia. Moreover few efforts have the size of amplified fragments depend on length and been made to assess the relative value of different sequence of short, single and arbitrary primers. methods for genetic characterization of tilapia germ- Priming sites are randomly distributed throughout a genome and polymorphisms in such sites result in plasm. The most recent classification of tilapiine species is differing amplification products, detected by the based on reproduction, development, feeding, struc- presence and absence of fragments. Such polymorph- tural characteristics and biogeography (Trewavas, isms are inherited in a Mendelian fashion and can be 1983). However, such characters are of limited value used as genetic markers. The method has been success- for identification purposes because they show con- fully used to detect variation between strains of siderable interpopulation variation and differences bacteria and rice (Welsh & McClelland, 1990), mice between species are small (Fryer & Iles, 1972; Abban, (Welsh et al., 1991), Gliricidia (Chalmers et al., 1992), 1988). Protein electrophoresis has been extensively closely related species of black Aspergilli (Megneg- used to discriminate species of tilapia (Kornfield at al., neau et al., 1993), and cocoa (Russell et a!., 1993) and 1979; McAndrew & Majumdar, 1983, 1984; Abban, between species of parasitic protozoa (Tibayrenc et a!., 1988; Sodsuk & McAndrew, 1991) and their hybrids 1993). (Macaranas et a!., 1986) but this technique could not This paper provides evidence that RAPD markers discriminate subspecies of 0. niloticus (Seyoum, 1989, can be used to discriminate between three widely culti- cited in Seyoum & Kornfield, 1992; Seyoum, 1990). vated species of tilapia and between several subspecies Mitochondrial DNA markers have been used success- of Oreochromis niloticus, probably the species with greatest commercial importance. An assessment is also made of the utility of RAPD analysis for systematic *Correspondence analysis of tilapia. 118 F. BARDAKCI & D. 0. F. SKIBINSKI

Materials and methods some modifications. Amplification reactions were per- formed in 500 ifiM KC1, 100 mvi Tris (pH 9.0 at 25°C), 1 per cent Triton X-100, 2.5 mtvi MgCl, 100 MM each of Species and subspecies stud/ed dATP, dTTP, dGTP and dCTP (Pharmacia), 5 pmoles Thespecies and subspecies used in this study are heldof 10-base primer, 0.5 units of Taq DNA polymerase in aquaria at the University College of Swansea, Wales. (Promega Biotec.), and 20 ng of genomic DNA in a Their sources are given in Table 1. 0. niloticus final volume of 25 ML. The mixture was overlaid with (baobab) is a commercial strain of 0. niloticus with the same volume of mineral oil. DNA amplification possible hybrid ancestry. It is not recognized as a sub- was performed in a thermal cycler (Hybaid, U.K.). For species but is designated as one for convenience in this the first cycle, denaturation, annealing and extension study. were 94°C for 2 mm, 35°C for 1 mm and 72°C for 2 Their sources are given in Table 1. 0. niloticus mm, respectively. Denaturation time was decreased to (baobab) is a commercial strain of 0. niloticus with 30 s for the following 44 cycles. One negative control possible hybrid ancestry. It is not recognized as a sub- (absence of template DNA) was performed for each set species but is designated as one for convenience in this of amplifications. study. Approximately 10 ML of amplification products were separated on 5 per cent vertical nondenaturing DNA extraction polyacrylamide gels in TBE buffer (0.89 M Tris, 0.89 M boric acid and 0.11 M EDTA, pH 8.3). Gels were fixed DNAwas prepared from fin tissue following the with 10 per cent ethanol and 0.5 per cent acetic acid method described by Hillis & Moritz (1990) with some solution twice for 3 mm, stained with 0.1 per cent silver modifications. Approximately 50 mg of the caudal fin tissue was cut into small pieces and suspended in 500 jcL STE (0.1 M NaC1, 0.05 M Tris and 0.01 M EDTA, Table 2 Sequence and operon codes of the random primers pH 8). After adding 30 ,uL SDS (10 per cent) and 30 used to study variation in Oreochromis species ML proteinase K (10 mg mL 1), the mixture was Primer codes incubated at 50°C for 30 mm. DNA was purified by Sequence (5'to3') successive extraction with phenol, phenol:chloro- 0PA03 AGTCAGCCAC form:isoamyl alcohol (25:24:1) and chloroform:isoamyl OPA 04 AATCGGGCTG alcohol (24:1), respectively. DNA was precipitated with OPA 05 AGGGGTCTTG ice-cold absolute ethanol and washed with 70 per cent OPA 07 GAAACGGGTG ethanol. The pellet was dried and resuspended in 150 OPA 08 GTGACGTAGG ,uL TE (10 mr'i Tris-HC1, 1 mtvi Na2EDTA.H20, pH OPA 10 GTGATCGCAG 7.2). OPA 12 TCGGCGATAG OPA 13 CAGCACCCAC OPA 17 GACCGCTTGT Amplificationconditions and electrophoresis OPA 19 CAAACGTCGG Aset of 13 decamer primers from Operon Technolo- OPB 08 GTCCACACGG OPC 02 GTGAGGCGTC gies was used in this study (Table 2). The amplification OPC 11 conditions were based on Williams et al. (1990) with AAAGCTGCGG

Table 1 Species and subspecies used in this study and their source

Species and subspecies Abbreviation Source

Oreochromis aureus Aur Lake Manzala, Egypt Oreochromis mossambicus Mos Aquarist stock Oreochromis niloticus 0. n. vulcani (vulcani) Vul Lake Vulcani, Kenya 0. n. baringoensis (baringo) Bar Lake Baringo, Kenya 0. n. niloticus (manzala) Nil Lake Manzala, Egypt 0.n. (baobab) Bao Baobab Fish Farm, Kenya

Common names of 0.niloticus subspecies are given in brackets. Nomenclature according to Trewavas (1983). RAPD ANALYSIS OF TILAPIA 119

nitratesolution for 10 mm, rinsed twice with distilled cnn0C') LL)r fiNfl LO0) W0) water, and then developed in an alkaline solution (1.5 040 (OLE) -cn (N per cent NaOH, 0.1 per cent NaBH4 and 0.15 per cent — r CH2O). A total of four individuals per species and sub- L11 B fl species (two males and two females) were analysed. Silver staining of polyacrylamide gels was found to be far superior to ethidium bromide staining of agarose 111 1I gels in terms of the number of fragments that could be III Ii resolved clearly. Analysis .3 TheRAPD patterns of individuals were compared within and between populations. Fragments were scored as 1 if present or 0 if absent. The index of similarity between individuals was calculated using the >Jflifl1 formula: :rm !IL np), where is the number of fragments shared by fill individuals x and y and n and n are the number of iirujj I fragments scored for each individual (Lynch, 1990). ) Within population similarity (S) is calculated as the Zfl:r.njI average of across all possible comparisons between individuals within a population. m. gjjp Between population similarity, corrected for within TffJJJ population similarity, is: a S= 1 + S— 0.5(S1+ S1), (00 Ofi cnN-C where S1 and S1 are the values of S for populations i (0 nfl LOLOO) 0) and j,respectivelyand S, is the average similarity N NO (ØLOfl N between randomly paired individuals from populations i and! (Lynch, 1990). It is possible for the value of S to I exceed 1. j H II S was also converted to a measure of genetic distance (D,1) using the equation: L-_HffrL1 D1 =— 1n[S/,J(S1S1)], (Lynch, 1991). D values were used to construct dendrograms using the unweighted pair-group method of analysis (UPOMA) (Sneath & Sokal, 1973). nfl The statistics can be calculated for each primer separately or for the combined data for all primers. F PP Results 'Tjp p. Allthe primers examined produced different RAPD fragment patterns (Fig. 1). The number of fragments generated per primer varied between six and 17. All =SWift primers gave species-specific RAPD patterns. Sub- z I Fig. I RAPD patterns from each of six populations using (a) up primer OPA 04 and (b) OPA 07. Samples (reading from the right hand lane): size markers; negative control (no template), t, ,fl 0. aureus, 0.mossambicus, 0. n. vulcani, 0. n. baringoensis, 0. n. niloticus, 0. n. (baobab). 120 F. BARDAKCI & D. 0. F. SKIBINSKI

0 species-specific patterns were obtained for some but CO 0 0 0 00.t- r) t rN 0 0O\ N - 0 'I00 C\ 0 not all primers. Values of similarity (S) between popu- zcood 000'0 0000 lations for each primer are given in Table 3. Two analyses were carried out on the data in Table 3 to 0 CO 00 L( I' N 0 00 0N N — \C assess the level of agreement between primers in the tf 0 -00\ N 00 N 0CO C\ N Coo0N 00 00 00 0 pattern of similarities and differences in population CO 00000 0 0oodd comparisons. In the first method of analysis, the corre- lation in identity between primers OPA 03 and OPA CO 0 N 0 N 00 v 0N ic N C t C00 C 0C flC NN 0000 00C\00N0000C0\ C\ N 00 00 04 across the 15 population comparisons was calcu- ooooddd lated. The calculation was repeated for primers OPA 03 and OPA 05 and so on for all possible primer pairs r —NN N C N LI V N N r N r) v — () 00C\ 0 of which there are 78 in total. Nineteen values were 00 C 0 00 C 00 N N 00 N C C 00 significant at the 1 per cent level and a further 12 signi- ficant at the 5 per cent level. All the significant values

V N NQ 0001) N r 00 —(•))— were positive and the average correlation was 0.482. In C N ) v0000 '-4 ) N 0000 — 0) NN h 0000N N — a factor analysis on the matrix of identity values, the 0 0000000 first principal component explained 45 per cent of the variance in identity. In the second method of analysis a N —'C)C N 0 N 00 m 00 C 0000 0 0 'C C 0 -NV two-way analysis of variance was carried out on the 4-0 C\C0000 Cl 0000 C 0 — C 0 00 C 0 values of Table 3 using the two-way interaction 0) between primers and population pairs as error. Signifi- C) N c'C N VN'C N N 'C N N 'C cant effects were observed both for primers (F12168 = Cl 'C 00 '- 'C 'C 0 'C C C ' 0 00NNQ0OON 9.9, Cl 0 C 0 d 00000000000 P<0.001) and population pairs (F14 168=11.4, C) P < 0.001). The results of the two methods of analysis suggest that the pattern of similarities and differences N (/)'C NCOLt N .- .-'- 0N 00 C — 'C' 00 CO N 'C 0 0000COO 00 NN C0 between populations showed broad agreement across '-C 0 C 0 d 0 0 0 0 00 C00 primers but that the overall level of similarity varied between primers. -C cC N 00 C 0 Q N - 0 O 0 'C '/ N N -'00 L(r 00oC- C\ N The highest value of within-population similarity (S) 0000 0 0000000000N00co- was obtained for 0. mossambicus, the lowest for 0. n. 4- C (baobab) (Fig. 2). The similarities between the subspe- Cl C) N 0N )r) r N0C N N 0 —r' cies of 0. niloticus (except between 0. n. vulcani and .0) " 0 -000 '1) 00 r N C\ c C 00 C) C0000OCOOCCN0COCOC 0. n. niloticus) are clearly higher than those between Cl 00 ——0000000 C) the species of Oreochromis (Fig. 3). The dendrogram (Fig. 4) links 0. n. mossambicus with the subspecies of C N N 00 L( 00 'C 0 'C N 0 N C) r') — C - N 0 N 'C L/) 00 N 0. niloticus, with 0. aureus as outgroup. To assess the 0) 00 N 'C CC LI) 0000 'C N 'C N 00 N N 0 significance of this grouping a compatibility analysis I) I- has been carried out by constructing separate dendro- C) N0L/) CO 'C '/) 00 If) 00 e COO N N grams for each of the 13 primers. Because there were C) 00 If) 0) —'C0 If) )) CO 0) 00 00d00doododN00NNN000000N four 0. niloticus subspecies, there were 4 x 13 52 C.) possible dendrograms involving 0. aureus, 0. mossam- CO C) — 0) 'C N 00 C) If) 'C 'C N 0 N C) 00 bicus and one of the 0. niloticus subspecies. Of these, 0000C) 0—N00N.NCSNN If) CN ) 000000000000 'C 'C N C) C) If) N the majority (a total of 25) had 0. aureus, rather than C)) another species, as outgroup. If the expectation that the three species are equally likely to be outgroup is If) r 'C 0 N If) NIf) r) Cl) —Cl)C) 00 N assumed as the null hypothesis, the expected number CO 'C c4) - 0 N— N—Cl)CO C) 0 Cl) N 0 'C00 If) C) N 00 N'C Cl) N 'C C) N 'C N E O000 of dendrograms with 0. aureus as outgroup is 52/ Cl 3 17.33 and =5.088 -u (P <0.05) and the null hypo- )C) 'C N Cl) thesis can be rejected. The favoured outgroup is thus CO 0 '--N If) Cl) CO N N N 'C N -C)f)N0N00NN-0 Cl) 0 N N 'C -,-ICCN 0 If) 0. aureus. Of the three subspecies (disregarding 0. n. Cl ood00000 C 00000 (baobab)), 0. n. baringoensis and 0. n. niloticus are Cl) If) N 000 N Cl) C) CO N — more closely linked with 0. n. vulcani as outgroup (Fig. ' — C) 00000-I-I00 4). Applying compatibility analysis, the number of indi- OOOOOCOOCOQ0)0/) vidual primer dendrograms with 0. iz. baringoensis, RAPD ANALYSIS OF TILAPIA 121

1 .C2

1.00

U) w C)2

(I)

Fig. 2 Average similarity (S) across .:: primers within species of Oreochromis 13 13 13 13 I) 13 and subspecies of 0. niloticus with 95 Aur Mos Vul Bar Nil Baa per cent confidence intervals. N = Number of pnmers SPECIES AND SUBSPECIES

11111 U)

.7 U, I .6

.5 Fig. 3 Similarity (Sq) across primers N= 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 between species of Oreochromis and AurtBao AurlMos AurNul Mos/Bar MosNul Bar/Nil VuWBao Vul/Nil Aur/Bar Aur,'NiI Mos/Bao Mos/Nil Bar/Baa Nil/Baa VulIBar subspecies of 0. niloticus with 95 per cent confidence intervals. N=number of primers COMPARISONS

Aur 0. n. vulcani and 0. n. niloticus as outgroup are zero, eight and five, respectively. Under the null hypothesis Mos that the three dendrograms should be equally frequent, Bar x=7.S44(P

I I I I I I I I 0.35 0.30 0.25 0.20 0.15 0.10 0.05 0.00 Discussion Incommon with other molecular techniques, RAPD GENETIC DISTANCE (D',1) analysis has been successful in identifying markers Fig.4 UPGMA dendrogram of the species of Oreochromis which distinguish tilapia species, although individual and the subspecies of 0. niloticus based on values of genetic primers differ significantly in the amount of inter- distance (D) calculated from data for all primers. population variation they detect. RAPD analysis has 122 F. BARDAKCI & D. 0. F. SKIBINSKI

also identified DNA markers which can distinguish at several regions in the genome, provides an advantage subspecies of Oreochromis niloticus and in this respect over other techniques. Even if some primers amplify has been more successful than allozyme analysis identical regions of the genome or if the technique itself (Seyoum, 1989, cited in Seyoum & Kornfield, 1992; is noisy, it should be possible to build up quickly a con- Seyoum, 1990). Mitochondrial DNA analysis has also sensus from patterns of interpopulation variation. revealed differences between subspecies of 0. niloticus (Seyoum & Kornfield, 1992) but might have some dis- advantages in studies of population differentiation Acknowledgements compared with nuclear markers. This is because of Wethank K.A. Naish for advice on computer analysis. evidence that mtDNA gene flow can occur across F. Bardakci is supported by the University of hybrid zones, in the absence of nuclear gene flow, in Cumhuriyet, Turkey. This work was funded in part by some (Ferris et al., 1983; Tegelstrom, 1987). grants from the Overseas Development Administration The tilapia species studied here hybridize readily in of the U.K. to J.A. Beardmore, G.C. Mair and D.O.E captivity, therefore the opportunity for mtDNA intro- Skibinski. gression is great. Intrapopulation RAPD variation was detected with all primers. This suggests that RAPD analysis might be References more sensitive than mtDNA analysis which has failed ABBAN,E. K. 1988. and Biochemical Genetics of to reveal variation within tilapia populations (Capili, some African Freshwater Fish Species. Ph.D. Thesis, 1990; Seyoum & Kornfield, 1992). In the present University of Wales, study, levels of intrapopulation variation are expected CAPILI, 3. B. 1990. Isozyme and Mitochondria! DNA Restriction to be low because most of the subspecies used were Endonuclease Analysis of Three Strains of 0. niloticus. founded, and have been propagated, from relatively Dissertation, University of Wales. small numbers of individuals. The population of 0. CHALMERS,K. J.,WAUGH, R., SPRENT.3. 1.,SIMONS, A. J. ANDPOWELL, mossambicus is descended from a single sib-mating w. 1992. Detection of genetic variation between and within and has the lowest level of variation. populations of Gliricidia sepium and G. inaculata using In the widely accepted classification of tilapia of RAPD markers. Heredity, 69,465—472. Trewavas (1983), 0. aureus and 0. niloticus are placed DEVO5, K. M. AND GALE, M. D. 1992. The use of random ampli- fied polymorphic DNA markers in wheat. Theor. App!. in a different subgenus from 0. mossambicus. The allo- Genet., 84, 567—572. zyme studies of McAndrew & Majumdar (1984) and FERRIS, 5. D., SAGE, R. D., HUANG, C. M., NIELSEN, 3. T., RITFE, U., Sodsuk & McAndrew (1991), favoured either this WILSoN, A. C. 1983. Flow of mitochondrial DNA across a classification or the closer linking of 0. mossambicus species boundary. Proc. Nat!. Acad. Sci. U.S.A., 80, with 0. aureus. The RAPD results are at variance with 2290—2294. these studies in suggesting a closer relationship FRYER,G. AND ILES, T. D. 1972. The Fishes of the Great between 0. mossambicus and 0. niloticus. In the Lakes of Africa. Oliver and Boyd, Edinburgh. mtDNA study of Seyoum & Kornfield (1992), 0. n. HADRYS, H., BALICK, M. AND SCHIERWATER, B. 1992.Application niloticus was more similar to 0. n. vulcani than to 0. of random amplified polymorphic DNA (RAPD) in molecular ecology. Mo!. Eco!., 1,55—63. n. baringoensis. The compatibility analysis gives least HILLIS, D. M. AND MORITZ, c. 1990. Molecular Systematics. support of all to this hypothesis. The evidence of highly Sinauer Associates, Sunderland, MA. significant differences between population pair similar- KAEMMER,D., AFZA, R., WEISING,K.,KAHL, G.AND NOVAK, F. j.1992. ity values provides support for the reliability of the Oligonucleotide and amplification fingerprinting of wild- technique as a tool for detecting genetic differences species and cultivars of banana (Musa spp.). BioI between populations. Technology, 10,30—35. RAPD analysis has been used for constructing trees KORNFIELD,I. L., RITTE, U.,RICHLER,C. AND WAHRMAN, j. 1979. in other organisms (Chalmers et al., 1992; Kaemmer et Biochemical and cytological differentiation among cichlid al., 1992; Megnegneau et al., 1993; Tibayrenc et al., fishes of the . Evolution, 33, 1—14. 1993). We have observed that small alterations in PCR LYNcH, M. 1990. The similarity index and DNA fingerprinting. parameters or quality of target DNA can radically alter Mo!. Bio!. Evol., 7, 478—484. RAPD patterns (see also, Devos & Gale, 1992; LYNCH, M. 1991. Analysis of population genetic structure by Williams et a!., 1993). Thus there may be reason to DNA fingerprinting. In: Burke, T., DoIf, G., Jeifreys, AJ. and Wolf, R. (eds) DNA Fingerprinting Approaches and view with caution systematic conclusions based on App!ications, pp. 113—126. Basel, Switzerland. RAPD analysis alone. On the other hand, the possibi- McANDREW, B. J. AND MAJUMDAR, K. C. 1983. Tilapia stock identi- lity of carrying out compatibility analysis with fication using electrophoretic markers. Aquacu!ture, 30, unlimited numbers of primers, each detecting variation 249—261. RAPD ANALYSIS OF TILAPIA 123

MCANDREW, B. J. AND MAJUMDAR, K. C. 1984. Evolutionary SODSUK, P. AND McANDREW, B. .i. 1991. Molecular systematics of relationships within three Tilapiine genera (Pisces: Cichli- three tilapiine genera Tilapia, and dae). Zoo!. .1. Linn. Soc., 80, 421—435. Oreochromis using allozyme data. J. Fish Biol., 39, MACARANAS, J. M., TANIGUCHL N., PANTE, M. J. R., CAPILI, J. B. AND 301—308. PULLIN, R. s. v. 1986. Electrophoretic evidence for extensive TEGELSTROM, H. 1987. Transfer of mitochondrial DNA from hybrid gene introgression into commercial 0. niloticus (L.) the northern red-backed vole (Clethrionomys rutilus) to stocks in the Philippines. Aqua. Fish. Manag., 17, the bank vole (C. glareolus). J. Mol. Evol., 24,218—227. 249—268. TIBAYRENC, M., NEUBAUER, K., BARNABE, C., GUERRINI, F., SKARECKY, MEGNEGNEAU, B., DEBETS, F. AND HOEXTRA, R. F. 1993. Genetic Ii AND AYALA, F. j.1993.Genetic characterization of six variability and relatedness in the complex group of black parasitic protozoa: Parity between random-primer DNA Aspergilli based on random amplification of polymorphic typing and multilocus enzyme electrophoresis. Proc. Nat!. DNA. Curr. Genet., 23, 323—329. Acad. Sci. U.S.A., 90,1335—1339. PULLIN, R. S. V. AND CAPILI, J. B. 1989. Genetic improvement of TREWAvAS, E. 1983. Tilapiine fishes of the genera Oreochro- : problems and prospects. In: Pullin, R. S. V., mis,Sarotherodon and Danakilia. British Museum of Bhukaswan, T., Tanguthai, K. and Maclean, J. L. (eds) The Natural History, London. SecondInternationalSymposium on Tilapia in Aquacul- WELSH, S. AND McCLELLAND, M. 1990. Fingerprinting genomes ture; ICLARM Conference Proceedings 15, pp. 259—266. using PCR with arbitrary primers. Nucleic Acids Res., 18, Manila. 7213—7218. RUSSELL, J. R., HOSEIN, F., JOHNSON, E, WAUGH, R. AND POWELL, W. WELSH, J., PETERSON, C. AND McCLELLAND, M. 1991. Poly- 1993. Genetic differentiation of cocoa (Theobroma cacao morphisms generated by arbitrarily primed PCR in the L.) populations revealed by RAPD analysis. Mo!. Ecol., 2, mouse: application to strain identification and genetic 89—97. mapping. Nucleic Acids Res., 19,303—306. SEYOUM, s. 1990. Allozyme variation in subspecies of WILLIAMS, S. G. K., HANAFEY, M. K., RAFALSKI, J. A. AND TINGEY, S. V. Oreochromis niloticus. Isozyme Bull., 23, 97. 1993. Genetic analysis using random amplified poly- SEYOUM, S. AND KORNFIELD, i. 1992. Identification of the sub- morphic DNA markers. Methods Enzymol., 218, species of Oreochromis niloticus (Pisces: Cichlidae) using 704—740. restriction endonuclease analysis of mitochondrial DNA. WILLIAMS, S. G. K., KUBELIK, A. R., LIVAK, K. J., RAFALSKI, J. A. AND Aquaculture, 102, 29—42. TINGEY, s. v. 1990. DNA polymorphisms amplified by SNEATH, P. H. A. AND SOKAL, R, R. 1973. Numerical Taxonomy. arbitrary primers are useful as genetic markers. Nucleic W.H. Freeman, San Francisco. Acids Res., 18,6531—6535.