Aquaculture, 33 (1983) 23-32 Elsevier Science Publishers B.V., Amsterdam - Printed in The Netherlands
23
DIFFERENCES IN THE AMOUNT AND DISTRIBUTION OF GENETIC VARIATION BETWEEN NATURAL POPULATIONS AND HATCHERY STOCKS OF ATLANTIC SALMON
GUNNAR STAHL Department
of Genetics,
University
of Stockholm,
S-l 06 91 Stockholm
(Sweden)
(Accepted 1 December 1982)
ABSTRACT G., 1983. Differences in the amount and distribution of genetic variation between natural populations and hatchery stocks of Atlantic salmon. Aquaculture,
St&l,
33: 23-32.
Tissue samples from a total of 1643 Atlantic salmon collected from naturally reproducing populations and hatchery stocks representing nine major river systems draining into the Baltic Sea were electrophoretically analyzed for 37 enzyme loci. Pronounced spatial genetic heterogeneity was generally observed among samples within as well as between different river systems. Samples representing hatchery stocks exhibit a significantly lower amount of genetic variability than natural populations. This is displayed as both a reduced variation within hatchery stocks and a lower amount of genetic divergence between them.
INTRODUCTION
Man’s impact on the Atlantic salmon (Salmo s&r) throughout its native distribution area has resulted in a drastic decrease in the total amount of naturally produced fish. In order to compensate for this reduction artificial propagation followed by a massive release of hatchery reared fish is frequently practiced. The Baltic Sea provides a good example; although the total amount of smelt produced yearly is almost the same today as it was in the beginning of the century, more than 60% is currently artificially reared (Christensen and Larsson, 1979; Johansson, 1981). The artificial propagation implies an obvious risk of reducing the total amount of genetic diversity within the species (Ryman and St&l, 1980, 1981; Ryman, 1981a; Allendorf and Phelps, 1980, 1981a; Cross and King, 1983, this volume). In order to avoid an unwanted reduction of genetic variation due to unwise management, a detailed knowledge both of the population structure and the amount and distribution of genetic variation throughout the species range is necessary. The examination of a large number of loci is required for a detailed
0044-8486/83/$03.00
o 1983 Elsevier Science Publishers B.V.
Torne Torne Torne Torne
Kahx KaIix Kahx KaIix Kalix
A
B
$ Luleiilven 2s
Byske $ Byske 1s
$ SkellefteZiven 1s
Wgde
$ IndaIs%lven 1s $ Indalsiilven 2s
$ Saimaa 1s
$ EmPn 2s
Byske iilv Byske iilv
SkeiIeftellven
Liigde g;lv
Indalsiilven IndaIsZlven
Saimaa
EmKn
Kaitum Ange&n Satter Ange&n Vettasjoki Ange&n Vaitiojoki
KdiX
Lainio Torne $ KukkoIa 2s $ Torne mynning 1s
Sample
Lule tiv
iilv Zlv glv iilv alv
Plv iilv iilv iilv
Drainage
MaP code
1981
1980
1980 1980
1979
1980
1979 1980
1981
1979 1979 1981 1981 1981
1979 1979 1980 1980
Year of sampling
82
63
162 103
69
390
59 40
31
98 98 46 79 29
14 100 50 130
Number of fish
0.500
0.912
0.923 0.937
0.986
0.890
0.890 1.000
0.855
0.622 0.622 0.783 0.627 0.638
0.720 0.640 0.719
0.429
Aat100
Locus
1.000
1.000
1.000 1.000
1.000
1.000
1.000 1.000
0.984
1.000 1.000 1.000 1.000 1.000
1.000 1.000 1.000 1.000
Agp-2 100
1.000
1.000
1.000 1.000
1.000
1.000
0.992 1.000
1.000
0.995 1.000 1 .OOO 0.981 1 .OOO
1.000 1.000 0.970 1.000
1.000
0.992
0.923 0.976
0.819
0.927
0.720 0.913
0.903
0.918 0.929 0.859 0.905 0.845
0.929 0.950 0.980 0.950
Mdh-3 Me-2 100 100
1.000
1.000
1.000 1.000
1.000
1.000
0.856 1.000
1.000
0.980 1.000 1.000 1.000 1.000
1.000 1.000 0.960 1.000
0.349
0.742
0.741 0.654
0.552
0.698
0.677 0.500
0.622
0.606 0.553 0.737 0.819 0.719
0.756 0.781 0.548 0.551
Pgm-1 Sdh-1 100 100
1.51
2.58
1.81 1.67
2.22
2.04
3.51 1.78
2.50
3.10 2.96 2.62 2.63 3.05
2.68 2.27 3.06 2.69
H% (37 loci)
Frequency of the most common allele (I 00) at six polymorphic loci and estimates of average heterozygosity (H) in samples representing naturally and artificially produced Atlantic salmon. The estimates of average heterozygosity were based on a total of 37 loci. For samples representing hatchery stocks ($) the year class is also specified (1s is one summer old fish ; 2s is two summers old fish)
TABLE I
2
25
analysis of population structure and for quantitative estimates of the distribution of genetic variation (Lewontin, 1974; Nei and Roychoudhury, 1974; Nei, 1975; Allendorf and Utter, 1979). A number of loci suitable for electrophoretic analysis has previously been described for the Atlantic salmon (Cross and Ward, 1980; St&l, 1981). An examination of 45 loci in naturally reproducing Atlantic salmon from northern Sweden revealed spatial genetic heterogeneity on a much finer geographical scale than was previously recognized (St&l, 1981). The aim of the present study is to estimate the amount and distribution of genetic variation within and between samples from both naturally reproducing populations and from hatchery stocks of Atlantic salmon native to river systems draining into the Baltic Sea. MATERIALS AND METHODS
Tissue samples of eye, liver, and muscle from a total of 1643 specimens representing nine major drainage systems were collected during the years 1979 to 1981 (Table I). The map code for each drainage (A-I, Table I) refers to the map of Scandinavia (Fig. 1) showing the geographical location of drainages sampled. Of the total number of 18 samples, there are nine samples representing hatchery stocks (marked with a “$“) and nine samples assumed to represent naturally reproduced fish. Offspring produced from a particular hatchery stock is based on yearly catches of mature fish that are migrating upstream to spawn. Consecutive yearclasses of offspring are therefore produced by different sets of parental fish. Each sample from a hatchery stock (Table I) represents a single yearclass of one (1s) or two summers (2s) old fish. Field collections were obtained through electrofishing a restricted geographical area within each drainage. Fish length ranged from 5 cm to 20 cm which indicates that mainly 2- and 3-year-old parr were sampled (0. Karlstrom, personal communication, 1980). No planting has taken place during the last years in any of the sampling areas and these samples should therefore represent the result of natural reproduction. The tissue samples were immediately frozen and transported to the laboratory where they were stored at - 60°C until electrophoretically analyzed. Electrophoretic techniques, staining procedures, interpretation of electrophoretic banding patterns, and designation of loci are according to Utter et al. (1974), Allendorf et al. (1977), Cross and Ward (1980), and StPhl (1981), respectively. Due to insufficient enzyme activity in some of the samples the total number of loci scored varied between 37 and 45. However, all fish were analyzed for the following 19 enzymes and the corresponding 37 loci screened for are given in parentheses (the system of nomenclature follows that of Allendorf and Utter (1979)): aspartate aminotransferase (Aat-I, 2, 3), alcohol dehydrogenase (A&), ol-glycerophosphate dehydrogenase
26
300
km
Fig. 1. Map of Scandinavia showing the geographical distribution of drainages that were sampled. The map code (A-I) refers to Table I.
(Agp-I), adenylate kinase (Ak-3), creatine phosphokinase (Cplz-I, 3), diaphorase @a), glyceraldehyde-3-phosphate dehydrogenase (Gap&-2), glutamate dehydrogenase (Gdh), p-glucoronidase (Gus), isocitrate dehydrogenase (Idh-I, 2, 3), lactate dehydrogenase (L&z-l, 2, 3, 4, 5), malate dehydrogenase (Mdh-I, 2, 3), malic enzyme (Me-l, 2, 3), 6-phosphogluconate dehydrogenase (d-Pgdh-I, 2), phosphoglucose isomerase (Pgi-I, 2), phosphoglucomutase (Pgm-I, 2), sorbitol dehydrogenase (Sdh-I, 2), superoxide dismutase (Sod) and xanthine dehydrogenase (Xdh). RESULTS
Genetic variation was observed at the following six loci; the segregating alleles are given in parenthesis: Aat(100, 50, and 25), Agp-2 (100 and 50), Mdh-3 (100, 115, and 80), Me-2 (100 and 125), Pgm-1 (100 and 75),
27
and Sdh-1 (100 and -50). Allele frequencies, given in Table I, were calfor all loci except for Sdh-1 where the freculated by “allele counting” quency of the 100 allele was computed from the square root of the proportion of Sdh-1 (100/l 00) homozygotes. There is an overall genetic heterogeneity between all 18 samples. Statistically significant allele frequency differences were obtained by contingency chi-square analyses at five out of the six genetically variable loci (A&3, x&, = 1624, P < 0.001; Me-2, xf, = 140, P < 0.001; Mdh-3, x:4 = 124, P < 0.001; Pgm-1, XT, = 331, P < 0.001; Sdh-I, XT, = 114, P < 0.001). Part of this heterogeneity was due to differences between samples within drainages (Table II). Multiple samples from the same drainage were available from four different drainages. Statistically significant allele frequency differences were obtained within all four drainages analyzed. The intrasample genotypic distributions were analyzed by comparing the observed genotypic distribution at codominant loci with the expected Hardy-Weinberg proportions calculated using Levene’s (1949) correction for limited sample size. A statistically significant excess of heterozygotes was observed in the sample $Em&, where all fish were heterozygotes at Aat- (XT = 81.00, P < 0.001). The genetic composition within samples was further analyzed by linkage disequilibrium tests between pairwise combinations of polymorphic loci (Hill, 1974). A significant deviation from gametic phase equilibrium was observed in three samples, two of which represented naturally reproducing populations (Anges&-Vettasjoki: Aat-s/Me-2, xi = 4.94, P < 0.05; Liigde: Aat-3/Me-2, x f = 4.59, P < 0.05) and one sample representing a hatchery derivative ($Em&: Aat-3/Sdh-I, xf = 17.67, P < 0.001). A total of 119 tests for Hardy-Weinberg proportions and linkage disequilibrium between polymorphic loci within samples were performed implying that about six tests are expected to result in statistical significance by pure chance. Among the four tests resulting in a statistical significance, the two samples Anges&r-Vettasjoki and Ltigde exhibited significance in just one test and at the 5% level only. Those significances were assumed to reflect type I statistical errors. In the sample $Em& two tests resulted in statistical significance and the level of significance was below 1% in both cases indicating true deviation from random genotypic distribution. The amount of genetic variation as measured by average heterozygosity (H) and based on the information at all 37 loci scored, ranged from 1.51% in $Em& to 3.51% in Byske (Table I). The unweighted average H for the nine samples representing naturally reproduced fish and the nine samples representing hatchery stocks are 2.8% and 2.2% (Table III), respectively. There is an overall reduction of more than 20% of the average heterozygosity within the hatchery stocks as compared to the natural populations, and the differences is statistically significant (P < 0.05 as revealed by Wilcoxon’s rank-sum test; Bradley, 1968).
3 4 1 1
4 5 2 2
A B D G
adf=8.
Torne iilv KaIiig;lv Byske iilv IndaIsiilven
df
Number of samples
Drainage
14.72** 6.90 9.76**
X2
X2
11.88** 257.24**=+ 7.72** 0.19
Mdh-3
Aot-3
2.14 6.36 0.68 5.64*
XZ
Me-2
X’
25.85*** 28.56*** 3.55 3.29
19.65*** 10.34* 10.84***
Sdh-1 XZ
Pgm-1
15 24 5 3
df
74.24*** 309.40*** 32.55*** 9.12*
x1
Total
Genetic homogeneity tests between samples within drainages by chi-square contingency comparisons of allele and phenotype counts at variable loci. Levels of significance are indicated by; * P < 0.05; ** P < 0.01; *** P < 0.001
TABLE II
29 TABLE III Gene diversity analysis (Nei, 1975) based on 37 enzyme loci in artificially and naturally produced Atlantic salmon from the Baltic Sea
Number of drainages sampled Total number of samples Average gene diversity within subpopulations (Ha) Gene diversity between subpopulations (DsT) Total gene diversity (HT=Ks+F~T)
Naturally reproduced
Artificially reproduced
4 9
7 9
0.028
0.022
0.0033
0.0025
0.031
0.024
DISCUSSION
There is a pronounced genetic heterogeneity among all 18 samples which were derived from nine different river systems draining into the Baltic Sea. The allele frequency differences obtained between samples which represent naturally produced Atlantic salmon from the same drainage (Table II) suggest several genetically distinct reproductive units within major drainages. As the present results are based on a single sample of parr from each locality there is a possibility that the allele frequency estimates in some samples represent a restricted number of parents (Allendorf and Phelps, 1981b). However, all field collections in the present study represent at least 2-year classes of parr and the observed genotypic distributions within samples are all in accordance with the expected distributions in large panmictic populations. In other salmonids the temporal variability of allele frequency estimates has shown to be of minor importance in relation to observed spatial allele frequency differences (Altukhov et al., 1975; Allendorf and Utter, 1979; Grant et al., 1980; Ryman, 1983, this volume). In view of the general tendency for salmonids to evolve genetically discrete and ecologically specialized subpopulations (Behnke, 1972; Allendorf, 1975; Altukhov et al., 1975; Ryman et al., 1979; Riddell et al., 1981; Ryman, 1981a, 1983, this volume; Ryman and St&l, 1981), the present observed spatial genetic heterogeneity in Atlantic salmon indicate that “river” should not constitute the basic unit when identifying discrete stocks; the basic unit should rather be the local subpopulation within the different rivers (cf., Mtiller, 1970). Allele frequency differences were observed between samples representing different derivatives of the same hatchery stock (rivers Torne alv and Indals&en, Table II). One possible explanation of the heterogeneities within hatchery stocks may be the use of a small effective number of parents when maintaining the stocks resulting in allele frequency changes due
30
to random genetic drift (Ryman and St&l, 1981). The extreme genotypic distribution observed in the sample $Em%l, where all individuals were heterozygous at the Aatlocus, actually suggests that a very restricted number of males and/or females (homozygous for different alleles at Aat-3) was used to produce that group of fish. The natural subdivision of Atlantic salmon into several genetically differentiated subpopulations within a river system might also explain the observed genetic heterogeneities between samples from the same hatchery stock since when collecting spawners, fish belonging to different genetically discrete subpopulations of a particular river may unintentionally be mixed when founding or propagating the stock. The amount of genetic variation allocated within and between samples was estimated by a gene diversity analysis (Nei, 1975; Chakraborty et al., 1982). A more than 20% reduction of average heterozygosity (H) or mean expected panmictic diversity (Hs, Table III) in samples representing artificially reproduced fish compared to naturally reproduced fish, suggests that a considerable loss of genetic variation is associated with hatchery operations. As a matter of fact, the loss of genetic variation within hatchery stocks may proceed at a considerable higher rate than indicated by the present estimate. As discussed above, there are indications that several hatchery stocks may constitute a mixture of two or more genetically distinct natural populations, and the gene diversity within such stocks is expected to be higher than what is typically found in natural populations. The present observations stress the immediate need for detailed analyses of the reasons for the apparent genetic erosion within hatchery stocks. Another important observation (Table III) is that of a reduced amount of differentiation between stocks as measured by &jT, which is approximately 25% lower than the differentiation observed among natural populations. Thus, hatchery stocks appear to be genetically more similar to one another than what is typical for natural populations. Altogether, the total gene diversity (&) is approximately 23% lower for the hatchery stocks, and this difference is due to a reduction of both the amount of variation within stocks and the amount of divergence between them. This is an alarming observation, in particular considering the fact that the hatchery stocks are assumed to represent a somewhat larger geographical area than the natural populations. Therefore, the amount of genetic divergence among the hatchery stocks should not be smaller than that among natural populations, provided that the existing genetic diversity had been adequately sampled and conserved. The present results point to the existence of a more heterogeneous genetical population structure of the Atlantic salmon in the Baltic Sea than has previously been aknowledged. The observation of a significantly lower amount of genetic variation in hatchery stocks as compared to natural populations indicate that currently practiced hatchery procedures may be associated with genetic erosion progressing at an alarming rate.
31
There is a strong need for the formulation of a strategy how to best conserve the genetic resources that are still available (cf., Ryman, 1981b). In both natural salmonid populations and hatchery stocks, electrophoresis has proven a valuable tool to characterize genetic variability patterns warranting its use in the planning and implementation of management programs. ACKNOWLEDGEMENTS
I would like to express my sincere thanks to N. Johansson and 6. Karlstrom for their aid with the collection of samples and to N. Ryman, U. Gyllensten, and 0. Leimar for valuable comments on this paper. The investigation was supported by grants from the Swedish National Environment Protection Board and the Swedish Natural Science Research Council. REFERENCES Allendorf, F.W., 1975. Genetic variability in a species possessing extensive gene duplication: Genetic interpretation of duplicate loci and examination of genetic variation in populations of rainbow trout. Ph.D. Thesis, University of,Washington, Seattle, WA. Allendorf, F.W. and Phelps, S.R., 1980. Loss of genetic variation in a hatchery stock of cutthroat trout. Trans. Am. Fish. Sot., 109: 537-543. Allendorf, F.W. and Phelps, S.R., 1981a. Isozymes and the preservation of genetic variation in salmonid fishes. In: N. Ryman (Editor), Fish Gene Pools. Ecol. Bull. (Stockholm), 34: 37-52. Allendorf, F.W. and Phelps, S.R., 1981b. Use of allele frequencies to describe population structure. Can. J. Fish. Aquat. Sci., 38: 1507-1514. Allendorf, F.W. and Utter, F.M., 1979. Population genetics. In: W.S. Hoar, D.J. Randall, and J.R. Brett (Editors), Fish Physiology, Vol. 3. Academic Press, New York, pp. 407-454. Allendorf, F.W., Mitchell, N., Ryman, N. and St&l, G., 1977. Isozyme loci in brown trout (S&no trutta L.): detection and interpretation from population data. Hereditas, 86: 179-190. Altukhov, Y.P., Salmenkova, E.A., Konovalov, S.M. and Pudovkin, A.I., 1975. Stationary distributions of the frequencies of lactate dehydrogenase and phosphoglucomutase gene in a system of subpopulations of a local fiih stock of Oncorhynchuo nerka Walb. I. Stability of a stock over generations with simultaneous variability of the component subpopulations. Genetica, 11: 44-53. Behnke, R.J., 1972. The systematics of salmonid fishes of recently glaciated lakes. J. Fish. Res. Board Can., 29: 639-671. Bradley, J.V., 1968. Distribution-free Statistical Tests. Prentice Hall, London, 388 pp. Chakraborty, R., Haag, M., Ryman, N. and StBhl, G., 1982. Hierarchical gene diversity analysis and its application to brown trout population data. Hereditas, 97: 17-21. Christensen, 0. and Larsson, P.-O., 1979. Review of Baltic Salmon Research. ICES Coop. Res. Rep., 1979 (89), 124 pp. Cross, T.F. and King, J., 1983. Genetic effects of hatchery rearing in Atlantic salmon. Aquaculture, 33 : 33--4O. Cross, T.F. and Ward, R.D., 1980. Protein variation and duplicate loci in the Atlantic salmon, Salmo salar L. Genet. Res. Camb., 36: 147-165.
32 Grant, W.S., Milner, G.B., Krasnowski, 0. and Utter, F.M., 1980. Use of biochemical genetic variants for identification of sockeye salmon (Oncorhynchus nerka) stocks in Cook Inlet, Alaska. Can. J. Fish. Aquat. Sci., 8: 1236-1247. Hill, W.G., 1974. Estimation of linkage disequilibrium in randomly mating populations. Heredity, 33: 229-239. Johansson, N., 1981. General problems in Atlantic salmon rearing in Sweden. In: N. Ryman (Editor), Fish Gene Pools. Ecol. Bull. (Stockholm), 34: 75-83. Levene, H., 1949. On a matching problem arising in genetics. Ann. Math. Statist., 20: 91-94. Lewontin, R.C., 1974. The Genetic Basis of Evolutionary Change. Columbia Univ. Press, New York, 346 pp. Mdller, D., 1970. Transferrin polymorphism in Atlantic salmon (Salmo solar). J. Fish. Res. Board Can., 27 : 1617-1625. Nei, M., 1975. Molecular Population Genetics and Evolution. North-Holland, New York, 288 pp. Nei, M. and Roychoudhury, A., 1974. Genetic variation within and between the three major races of man. Caucasoids, Negroids, and Mongoloids. Am. J. Hum. Genet., 26: 421-443. Riddell, B.E., Leggett, W.C. and Saunders, R.L., 1981. Evidence of adaptive polygenic variation between two populations of Atlantic salmon (Salmo salar) native to tributaries of the S.W. Miramichi River, N.B. Can. J. Fish. Aquat. Sci., 38: 321-333. Ryman, N., 1981a. Conservation of genetic resources: experiences from the brown trout (Salmo trutta). In: N. Ryman (Editor), Fish Gene Pools. Ecol. Bull. (Stockholm), 34: 61-74. Ryman, N. (Editor)., 1981b. Fish Gene Pools. Preservation of Genetic Resources in Relation to Wild Fish Stocks. Ecol. Bull. (Stockholm), 34,111 pp. Ryman, N., 1983. Patterns of distribution of biochemical genetic variation in salmonids: differences between species. Aquaculture, 33: 1-21. Ryman, N. and Stahl, G., 1980. Genetic changes in hatchery stocks of brown trout (Salmo trutta). Can. J. Fish. Aquat. Sci., 37: 82-87. Ryman, N. and St&h& G., 1981. Genetic perspectives of the identification and conservation of Scandinavian stocks of fish. Can. J. Fish. Aquat. Sci., 38: 1562-1575. Ryman, N., Allendorf, F.W. and Stlhl, G., 1979. Reproductive isolation with little genetic divergence in sympatric populations of brown trout (Salmo trutta). Genetica, 92: 247-262. StPhl, G., 1981. Genetic differentiation among natural populations of Atlantic salmon (Salmo salar) in northern Sweden. In: N. Ryman (Editor), Fish Gene Pools. Ecol. Bull. (Stockholm), 34: 95-105. Utter, F.M., Hodgins, H.O. and Allendorf, F.W., 1974. Biochemical genetic studies of fishes : potentialities and limitations. In: D.C. Malins and J.R. Sargent (Editors), Biochemical and Biophysical Perspectives in Marine Biology. Academic Press, New York, pp. 213-238.