Hereditas I1 7: 45-50 (1992)

Genetic variability within and among sympatric brown ( tvutta) populations: multi-locus DNA fingerprint analysis

PAUL0 A. PRODOHL, JOHN B. TAGGART and ANDREW FERGUSON

Division of Envirmmental and Evolutionary Biology, School of Biology and Biochemistry, The Queen’s University of Belfast, Bevast BT7 INN, Northern Ireland, UK

PROD~HL,P. A,, TACGART,J. B. and FERGUSON,A. 1992. Genetic Variability within and among sympatric (Salmo trulta) populations: multi-locus DNA fingerprint analysis. - Hereditas 117 45-50. Lund, Sweden. ISSN 0018-0661. Received November 8, 1991. Accepted March 30, 1992

Three brown trout (Salmo trutta L.) morphotypes, sonaghen, gillaroo, and ferox have been described from , Ireland. Previous isozyme studies have shown these to be three genetically distinct and reproductively isolated populations. Multi-locus DNA fingerprinting of these populations was carried out to provide an initial assessment of this technique’s potential use for salmonid population genetic studies. Highly variable profiles of Pal I digested genomic DNA were revealed for all three populations, using the human minisatellite 33.6 hybridization probe. Band-sharing analysis concurred with previous isozyme studies in showing a lower level of genetic variability within the ferox as compared with sonaghen and gillaroo. The analysis also confirmed the genetic distinctness of each population. While no unambiguous diagnostic fragments were observed for any of the populations, it was possible to discriminate ferox from both sonaghen and gillaroo based on the higher mean band-sharing coefficient observed within ferox. The main value for multi-locus DNA fingerprinting in salmonid population genetic studies may be as an alternative rapid method of determining overall levels of genetic variability (heterozygosity).

Paul0 A. Prodohl, Division of Environmental and Evolurionary Biology, School of Biology and Biochemistry, The Queen’s University of Belfast, Belfit BT7 INN, Northern Ireland, UK

Electrophoretic studies at a selection of protein- At the population level, however, this methodol- coding loci have proved valuable for obtaining ogy has proved to be less applicable, DNA finger- genetic information concerning salmonid stock prints generally being too variable to reliably infer structure, gene pool conservation, etc. (ALLEN- relatedness beyond the close family (BURKE1989; DORF and UTTER1979; RYMAN1983; FERGUSONLEWIN 1989). Nevertheless, there have been a few 1989). However, this approach has often been notable studies. For example, from comparisons of severely restricted by the relatively low levels of DNA fingerprint band-sharing coefficients, KUHN- electrophoretically detectable polymorphism ob- LEIN et al. (1989) calculated an index of genetic served among natural populations. Alternative distance among poultry strains. This was found to sources of highly polymorphic genetic markers reflect closely the known history and relationships would clearly be advantageous. of the different populations. Following a similar Of potential interest in this respect are the hy- approach, GILBERTet al. (1990) estimated relative pervariable non-coding regions of nuclear DNA, genetic variability within and distance among small known as ‘minisatellite’ or VNTR (variable num- island populations of California Channel Island ber tandem repeat) DNA, which are detected by foxes (Urocyon littoralis), the data being used to DNA fingerprinting methodologies (JEFFREYS et reconstruct their recent phylogeny. al. 1985a,b; VASSARTet al. 1987). Resultant re- Recently it has been demonstrated that the hu- striction fragment profiles are often highly variable man minisatellite probes 33.6 and 33.15 (JEFFREYS among individuals and usually individual specific. et al. 1985a) can be used to detect highly complex As such, they are indicative of multi-allelic poly- DNA fingerprints in salmonid fishes (TAGGART morphisms at many loci. Since the technique can and FERGUSON1990a). However, their potential be highly discriminatory, DNA fingerprinting has as population genetic markers has yet to be as- been widely exploited to assign parentage and sib- sessed. To this end, due to the availability of ship within social groups of (WETTONet extensive background genetic data, the brown al. 1987; BURKE 1989; AMOS et al. 1991; TEGEL- trout (Salmo trutta L.) populations of Lough STROM et al. 1991). Melvin, north-west Ireland provide an appropriate 46 P A. PRO~HLET AL. Hereditas J I7 (1992) benchmark study group. Three brown trout mor- FERCUSON( 1990a). However, different hybridiza- photypes, known locally as sonaghen, gillaroo, and tion conditions were employed; with prehybridiza- ferox, have been described from this lake. Exten- tion (4 h) and hybridization (16 h) being carried sive isozyme, ecological, and morphometrical stud- out in 3X SSPE (0.54 M NaCI, 0.03 M NaH,PO,, ies have confirmed that these three types are 3 mM EDTA Na, salt, pH 7.7), 6 YOpolyethylene genetically distinct, reproductively isolated popula- glycol 8000, 1 YO sodium dodecyl sulphate, 0.5 YO tions (summarised in FERGUSON1986; FERGUSONdried milk, at 65°C (DALGLEISH1987). and TAGGART1991). This paper presents the re- sults of a preliminary investigation of multi-locus DNA fingerprint analyses DNA fingerprinting of Melvin brown trout. All DNA fingerprinting analyses were based on visual inspection of banding patterns directly from autoradiographs. In addition to looking for diag- Materials and methods nostic minisatellite fragments among all 83 individ- Fish surnples uals, band-sharing coefficients were calculated. The simple similarity statistic S = 2N,,/( N, + Ny), It has been shown that the three Melvin brown where N, and N, are the number of fragments in trout types maintain their genetic integrity as a individual x and y respectively, and N,, is the result of their distinctive spawning habits (see FER- number shared by both, was used (LYNCH1990). GUSON and TAGGART for details). Thus, 1991 As the DNA fingerprints were particularly com- juvenile fish of each type were obtained by elec- plex, band analyses were restricted to the larger, trofishing, in September 1990, in three separate more clearly resolved fragment sizes, i.e., above rivers: Drowes (gillaroo, n = 30); Tullymore (son- 6.6 kb (kilobase pairs). Furthermore, S,, values aghen. n = 30); ( = 23). lower Glenaniff ferox, n were determined only from comparisons of adja- Care was taken to sample fish over as wide an area cent fingerprints (of comparable intensity and reso- as possible, and from at least two cohorts (0+ and lution) within the same gel; i.e., lane 1 with 2, 2 1 years), to reduce the likelihood of capturing + with 3, 3 with 4, etc. For intra-population com- siblings. The fish were immediately frozen on dry parisons the single best resolving autoradiograph ice and later stored at -4OC until required. (19 lane gel) for each morphotype was analysed. Confirmation of typing was provided by morpho- Combined S,, values derived from three separate logical examination and electrophoretic analysis of gels, each having mixed lanes of six sonaghen, six muscle. liver, eye, and brain tissues for known gillaroo, and six ferox individuals, were used for diagnostic enzyme encoding loci (TAGGARTet al. inter-population comparisons. 1981: FERGUSONand TAGGART1991). Limited Although S,, values should be approximately sonaghen spawning is also known to occur in the normal in distribution (LYNCH 1990), following Glenaniff river. Therefore, in order to preclude the recommendations of SOKALand ROHLF(1981) mistyping of ferox. only fish homozygous for the for frequency based data, analyses were performed diagnostic eye specific lactate dehydrogenase ances- on arc-sine transformed similarity values. With tral allele. i.e., Ldh-5( 100~100),were classified as each individual being used in a maximum of two ferox (FERGUSONand TAGGART1991). comparisons S,, values were considered to be suffi- ciently independent to apply the standard formula DNA fingerprinting for sampling variance of the mean population sim- ilarity (S). Such estimates of sampling variance, DNA fingerprinting was carried out using the hu- however, are likely to be slightly downwardly bi- man minisatellite 33.6 hybridization probe ( JEF- ased since the data are not completely independent FREYS et al. 1985a) on Pa/ I digested genomic (LYNCH 1990). The analyses were repeated on DNA. Due to its poorer resolution in brown trout much reduced, but fully independent sets, based on (TAGGARTand FERGUSON1990a), the 33.15 unique non-overlapping pairs of individuals as sug- probe was not employed in this study. DNA gested by LYNCH(1990). In all cases the absolute extraction (from liver tissue). digestion. electro- variance values were little different from those phoresis. Southern blotting, probe labelling, post- presented, and significance levels were unaffec- hybridization washing, and autoradiography, es- ted (data not shown). Approximate values of sentially followed that described by TAGGARTand WRIGHT’S( 1951) index of population subdivision, Hereditas 117 (1992) DNA FINGERPRINTINGOF BROWN rRour POPULATIONS 47

F,,, and NEI'S (1972) genetic distance, D, were calculated from the formulae derived by LYNCH (1991).

Results All three brown trout populations revealed vari- able multi-banded patterns with the 33.6 human minisatellite probe, Fig. 1. While no single restric- tion fragment was found to be unambiguously diagnostic for a particular population type, a pos- sible ferox specific band was noted (arrowed in Fig. 1). However, its specificity was not conclu- sively demonstrated since this region was not con- sistently resolved in all the Southern blots analy- sed. Several bands predominantly associated with only one or two populations were also evident. The results of band-sharing analyses, both within and between population groups, are sum- marized in Table 1. One way analysis of vari- ance (ANOVA) of arc-sine transformed data re- Fig. 1. DNA fingerprints of five ferox (FER), gillaroo vealed statistically significant differences (F2,46= (GILL) and sonaghen (SON). In each case, 3 pg of Pal I 70.6, P 0.1). In all Discussion three cases, however, these pairwise combinations were measurably more heterogeneous than both of The DNA fingerprints of the Melvin populations their constituents, intra-population S values being proved to be as variable and complex as those statistically significantly higher than inter-popula- previously reported for other brown trout (TAG- tion s values, Table 2. GART and FERGUSON1990a). However, despite FSTand D values based on fingerprint analyses this high level of complexity and the technical are given in Table 3, together with comparable limitations of fragment resolution, several aspects statistics calculated from extensive isozyme data of the genetic structure of the populations were for these same populations (from FERGUSONand revealed. TAGGART1991). While no firm conclusions can be Although not equivalent, the similarity index S drawn from this very limited comparison, it is, can be used as an approximate (upwardly bia- nevertheless, interesting to note the similarity be- sed) estimator of homozygosity (LYNCH 1990; tween the computed FSTvalues from both sources. GILBERTet al. 1991). The higher level of band- Furthermore, both genetic distance analyses show sharing among ferox individuals is thus likely to be sonaghen and gillaroo to be more similar to each indicative of a greater degree of genetic homogene- other than either is to ferox. As would be expected ity within ferox as compared with sonaghen and 48 P A. PRO~HLET AL Hereditas I I 7 (I 992)

ruble 1. Summary of band-sharing statistics: a) within; b) between the three brown trout types a) Intra-Population Comparisons

Sonaghen Gillaroo Ferox

No. pairwise comparisons 15 17 16 No. bands scored: Mean(SD) 28.4 (2.6) 27.8 (5.2) 30.0 (3.7) Band-sharing coeR. (s): Mean(SD) 0.45 (0.03) 0.45 (0.08) 0.68 (0.07)

b) Inter-Population Comparisons

Son,'Fer Son!G:ll FeriGill

No. pairwise comparisons 16 14 16 No. bands scored: Mean(SD) 26.6 (2.9) 25.1 (."A) 25.6 (3.4) Rand-sharing coeff. (s): Mean(SD) 0.35 (0.1) 0 3s (0.1) 0.30 (0.09)

I ohk -1. Comparison of each pairwise inter-population mean identified as ferox (FERCUSON and TAGCART hand-sharing coefficient (s)with its constituent intra-population 1991). S balues: one wak ANOVA and T' pairwise comparirons (under- lined). Computed from arc-sine lransformed data Fingerprint analysis may thus provide an alter- native method of estimating multi-locus hetero- ANOVA zygosity in populations, for example from a con- Sonagheii Son Fer Ferox F df P servation genetic perspective. In many cases this 0.45 0.15 0.68 71.4 2.44

TAGCART,J. B., HYNES, R. A., PROD~HL,P. A. and FERGUSON. VASSART, G., GEORGES, M., MONSIEUR,R., BROCAS, H., A. 1992. A simplified protocol for routine total DNA isolation LEQUARRE,A. S. and CHRISTOPHE,D. 1987. A sequence in MI3 from salmonid fishes. - J. Fish Bid. (in press) phage detects hypervariable minisatellites in human and TEGELSTROM,H., SEARLE,J., BROOKFIELD,J. and MERCER,S. DNA. -Science 235 683-684 1991. Multiple paternity in wild common shrews (Sorex araMus) WEITON,J. H., CARTER,R. E., PARKIN,D. T. and WALTERS,D. is confirmed by DNA fingerprinting. - Heredity 66: 373-379 1987. Demographic study of a wild house sparrow population TURNER,8. J., ELDER,J. F. and LAUGHLIN,T. F. 1991. Repetitive by DNA fingerprinting. - Nature 327: 147--149 DNA sequences and the divergence of fish populations: some WRIGHT,S. 1951. The genetical structure of populations. -Ann. hopeful beginnings. - J. Fish Bid. 39(suppelernenl A): 131 - 142 Eugen. 15: 323-354