Loss of genetic variation at microsatellite loci in hatchery produced abalone in Australia (Haliotis rubra) and South Africa (Haliotis midae)

Aquaculture 233 (2004) 109 – 127 www.elsevier.com/locate/aqua-online

Loss of genetic variation at microsatellite loci in hatchery produced abalone in Australia (Haliotis rubra) and South Africa (Haliotis midae) B. Evans a,b,c,d, J. Bartlett a, N. Sweijd e, P. Cook e, N.G. Elliott a,* a CSIRO Marine Research, GPO Box 1538, Hobart, Tasmania 7001, Australia School of Zoology, University of Tasmania, GPO Box 252-05, Hobart, Tasmania 7001, Australia c Aquaculture CRC Ltd. PO Box 123, Broadway, NSW 2007, Australia d School of Marine Biology and Aquaculture, James Cook University, Townsville, 4810, Australia e Department of Zoology, University of Cape Town, Private bag Rondebosch, 7701, Cape Town, South Africa b

Received 11 August 2003; received in revised form 24 September 2003; accepted 28 September 2003

Abstract Microsatellite DNA markers were used to investigate levels of genetic diversity within cultured populations of Haliotis midae and Haliotis rubra in South Africa and Australia, respectively. The cultured populations examined were F1 progeny of wild caught broodstock. All populations show a decline in genetic diversity, measured as the number of alleles per locus (35 – 62% allele loss) when compared to wild stocks in the area of respective broodstock collection. There was, however, no associated loss of heterozygosity. Changes in the frequency of alleles between farmed and wild samples were observed in both species. Mean levels of genetic relatedness for the cultured H. midae were not significantly different to zero, while those for the cultured H. rubra were significantly higher. The estimated effective population size of H. midae broodstock was between 75.3 (SD F 57.6) and 43.5 ( F 29.8) for a west coast farm and between 18.5 ( F 8.4) and 16.8 ( F 8.0) for an east coast farm. The observed loss of alleles in both farm samples was significantly greater than that expected due to genetic drift based on such effective population size estimates. The effective population size of a farm sample of H. rubra was estimated at between 27.2 ( F 3.8) and 22.4 ( F 4.7). The observed loss of alleles in this instance was not significantly greater than expected due to genetic drift. D 2004 Elsevier B.V. All rights reserved. Keywords: Genetic relatedness; DNA; Genetic diversity; Aquaculture; Heterozygosity; Genetics

* Corresponding author. Tel.: +61-3-6232-5263; fax: +61-3-6232-5090. E-mail address: [email protected] (N.G. Elliott). 0044-8486/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.aquaculture.2003.09.037

110

B. Evans et al. / Aquaculture 233 (2004) 109–127

1. Introduction Microsatellites are non-coding, highly polymorphic, co-dominant DNA markers, that are a powerful tool to measure the response of genetic variation in hatchery populations subjected to different breeding strategies, and for the identification of closely related individuals (Villanueva et al., 2002). The high fecundity (>1 million eggs per female) of mature abalone, and artificially high survival rate of juvenile abalone in a culture environment ensures that sufficient seed for each year’s production may result from only a small number of parents. When small numbers of broodstock are used, or the contribution of each parent is unbalanced, or related individuals are mated, there may be a decrease in genetic variability of farmed stocks (Boudry et al., 2002; Elliott and Reilly, 2003). Such problems are amplified when hatchery reared broodstock are used outside of a genetically controlled breeding program (Hahn, 1989). Loss of genetic variation at allozyme loci has been reported in hatchery populations of salmonids subject to multiple generations of culture (Ryman and Stahl, 1980; Vuorinen, 1984), and also in first generation hatchery populations of salmon (Verspoor, 1988), and several invertebrate species (Hedgecock and Sly, 1990; Hedgecock et al., 1992; Benzie and Williams, 1996). More recent microsatellite analysis of wild and farmed Atlantic salmon populations revealed a decrease in genetic variability of farmed stocks in terms of allelic diversity but not in overall heterozygosity (Norris et al., 1999). Previous studies of genetic diversity in first generation hatchery stocks of abalone have utilised allozyme electrophoresis (Smith and Conroy, 1992; Mgaya et al., 1995; Gaffney et al., 1996—but see Burton and Tegner, 2000). These studies all report the extinction of rare wild alleles from the farmed populations examined. Reductions in genetic variation have been shown to be detrimental to commercially important traits such as growth rate (Koehn et al., 1988) and fitness (Danzmann et al., 1989) in other marine organisms. Such reductions in abalone cultured for re-seeding operations could have a negative impact on native stocks in the area of stocking by reducing overall variation or swamping locally adapted genotypes (Allendorf and Ryman, 1988). This article is the first to employ microsatellite markers for the comparison of genetic variation in farmed and wild samples of abalone. Such changes in genetic variation may be measured as a shift in the frequency of common alleles, as a loss of rare alleles, and as a reduction in mean heterozygosity through loss of common alleles (Smith and Conroy, 1992). Here we present a comparison of variability at three microsatellite loci within and between samples of cultured and wild South African abalone (Haliotis midae). The wild samples are from the extent of the species range, and the two hatchery samples come from farms at the western and eastern regions of the range. Levels of genetic variation in wild samples of H. midae were examined previously (Sweijd, 1999; Evans et al., 2000, unpublished data), with significant differentiation observed between samples from either side of Cape Agulhas. We also present variation at five microsatellite loci within and between four cohorts of farmed Tasmanian blacklip abalone (Haliotis rubra) and compare these with that observed in wild samples from around Tasmania. Levels of genetic variation within natural stocks of Australian blacklip abalone are currently being investigated at eight microsatellite loci in samples from the species complete range in southern Australia (Elliott et al., 2003, unpublished).

B. Evans et al. / Aquaculture 233 (2004) 109–127

111

2. Methods 2.1. Samples H. midae: hatchery: Fifty H. midae individuals from a single year class were obtained from a commercial abalone farm on the south –west coast of South Africa (western farm). The sample consisted of a randomly selected group of 6-month-old juveniles that were obtained from a single growout tank, and represent progeny from multiple broodstock. The broodstock were originally obtained from a mixture of Western Cape wild populations. A further 50 H. midae individuals were collected from a commercial abalone farm on the east coast of South Africa (eastern farm). This sample comprised 6-month old juveniles from multiple broodstock obtained from Cape Recife on the east coast. All cultured abalone collected in South Africa were transported live from the farm to the University of Cape Town. They were then individually dissected using sterile instruments to isolate gill tissue, which was stored at 20 jC until DNA extraction. H. midae: wild: Commercial sized H. midae individuals were collected by SCUBA divers from six localities on the South African coast, three to the west (total 108) and three to the east (total 127) of Cape Agulhus a recognised area of genetic discontinuity (Sweijd, 1999; Evans et al., 2000, unpublished). H. rubra: hatchery: Sixty-four H. rubra juveniles were collected from each of four commercial tanks at a Tasmanian abalone farm. The individuals were all removed from a single randomly selected artificial substrate in each tank. Each tank is representative of an isolated cohort (separate multiple broodstock spawning) of individuals settled between 5 and 8 months earlier. Cultured H. rubra individuals were transported on ice to CSIRO Marine Research, Hobart, where they were individually dissected using sterile instruments to isolate gill tissue, which was then stored at 80 jC until DNA extraction. H. rubra: wild: Samples of H. rubra from seven Tasmanian sites (total 603), for which there was no evidence of genetic differentiation, were used in this study (Evans, 2002). All samples were collected by staff at the Tasmanian Aquaculture and Fisheries Institute, and were chilled on ice prior to dissection of gill tissue and subsequent storage at 80 jC. 2.2. DNA extraction Commercial DNA extraction kits (QIAGEN) were used to extract DNA from muscle tissue from South African cultured individuals. DNA was extracted from the gill tissue of Australian cultured samples and all wild individuals using a modified CTAB procedure (Grewe et al., 1993). 2.3. PCR amplification of microsatellite loci H. midae: Three microsatellite markers (CmrHr 2.15, CmrHr 2.23 and CmrHr 2.29: GenBank accession numbers; AF195956, AF302832, AF302834) were optimized for

112

B. Evans et al. / Aquaculture 233 (2004) 109–127

amplification in a Hybaid 96-well thermocycler. The development of microsatellite markers was reported previously (Evans et al., 2000). Amplification conditions consisted of an initial denaturation at 94 jC for 3 min followed by 30 cycles of denaturation at 94 jC (60 s), annealing at 51 jC (CmrHr 2.23) or 53 jC (CmrHr 2.15 and CmrHr 2.29) (45 s), and extension at 72 jC (60 s). Cycling was followed by a 6min extension step at 72 jC. Reactions were performed in a volume of 25 Al consisting of approximately 200 ng of genomic DNA template, 200 AM of each dNTP, 7 pM of each oligonucleotide primer, either 1.0 mM (CmrHr 2.23) or 3.0 mM (CmrHr 2.15 and CmrHr 2.29) MgCl2, two units of Taq DNA polymerase (Fisher Biotech) and sterile milli-Q water to volume. Two microlitres from each amplified microsatellite locus was mixed and diluted to a final volume of 80 Al. Two microlitres of this dilution was then mixed with formamide, loading dye and Genescan Tamra-500 size standard (ABI), denatured at 95 jC for two min, and loaded onto a 4% denaturing polyacrylamide gel. Samples were run on an ABI373 DNA autosequencer and genotypes determined with GenotyperR software. H. rubra: Five microsatellite markers were used to investigate H. rubra population differentiation and to investigate diversity in the four hatchery-produced cohorts. The markers used were: CmrHr 1.14, CmrHr 2.14, CmrHr 2.30, CmrHr 1.24 and RUBCA1 (GenBank accession numbers: AF195952, AF195957, AF195959, AF195953, AF027573). PCR amplifications were performed in 96-well plates (Costar) in a total volume of 25 Al. All loci were amplified in a multiplex reaction comprising 2.5 mM MgCl2, 0.2 mM each dNTP, 2.5 Al of 10  buffer (670 mM Tris – HCL pH 8.8, 166 mM (NH4)2SO4, 4.5% Triton X-100, 2 mg ml 1 gelatin), 0.66 units Taq DNA polymerase, 25 ng of template DNA and the reaction made up to 25 Al with sterile milli-Q water (all reagents from Fisher Biotech). The reaction comprised 3 pmoles of each CmrHr 1.14 primer, 4 pmol of each CmrHr 1.24 primer, 1 pmol of each CmrHr 2.14 primer, 4 pmol of each CmrHr 2.30 primer and 3 pmol of each RUBCA1 primer. All PCRs were conducted in a Perkin-Elmer 9600 thermal cycler. Cycling conditions comprised an initial denaturation step of 3 min at 94 jC and concluded with a final extension step of 10 min at 72 jC. Amplification involved 10 cycles of: denaturation at 94 jC for 30 s; annealing at 60– 55 jC for 30 s, dropping by 0.5 jC per cycle; and extension at 72 jC for 60 s. This was followed by a further 25 cycles of denaturation at 94 jC for 30 s; annealing at 55 jC for 30 s, and extension at 72 jC for 60 s. One microlitre from each amplification was diluted in 3 Al of sterile milli-Q water, and 0.7 Al of this dilution was mixed with 2 Al formamide, 0.5 Al loading dye and 0.5 Al Genescan Tamra-500 size standard (ABI), denatured for 2 min at 95 jC, and loaded onto a 4.8% denaturing polyacrylamide gel. Samples were run on an ABI-377 DNA autosequencer and genotypes determined using GenotyperR software. 2.4. Statistical analysis Genetic diversity for each locus per farm cohort was estimated, and compared to the diversity in wild samples of the same species. Diversity was measured by the number of alleles per locus and by the observed (Ho) and Hardy – Weinberg expected (He) heterozygosity. Ho, He, and tests for deviations from Hardy– Weinberg Equilibrium (HWE)

B. Evans et al. / Aquaculture 233 (2004) 109–127

113

within samples were estimated using GENEPOP Ver. 3.2 (Raymond and Rousset, 1995). An index of heterozygote deficiency or excess (D), where D=[Ho He]/He (Selander, 1970) was also calculated from the heterozygosity estimates. Significance levels for deviations from HWE were based on 100,000 steps of a Markov chain procedure. Linkage disequilibrium was assessed in each of the species groups using exact tests in GENEPOP Ver. 3.2 (Raymond and Rousset, 1995). Significance of departure from equilibrium levels was tested by a Markov chain procedure, with significance levels determined after 400 batches of 4000 iterations each. ARLEQUIN Ver. 2.000 (Schneider et al., 2000) was used for an analysis of variance of allele frequencies within and among samples, as well as between all sample pairs (AMOVA). The method is based on Excoffier (1993). ARLEQUIN also permitted multi-locus estimates of UST, an analogue of FST, the proportion of the total genetic variation attributable to population differentiation. Mean relatedness within and between cultured and wild samples of H. midae and H. rubra was estimated using the program Relatedness 5.0 (see Queller and Goodnight, 1989). Estimates were calculated against background allele frequencies obtained from 210 H. midae and 3558 H. rubra (Elliott et al., 2003 unpublished) individuals collected from natural populations around South Africa and Tasmania, respectively. Estimates of the effective population size (Nk) contributing to the hatchery stocks, and genetic drift based on these estimates were made from allele-frequency variances (Hedgecock and Sly, 1990). An adjusted chi-square goodness-of-fit test with one degree of freedom was used to test the significance of the observed loss of alleles compared to that expected due to genetic drift. The allele frequencies for a farm sample were compared with the combined wild samples for each respective area, and with a single collection. Therefore for H. midae, the allele frequencies for the west farm sample were compared with the combined frequencies of three collections from the west of Cape Agulhus, and then only with frequencies from a collection from the Kleinmond area. The east coast farm sample was compared with a combined sample from three collections to the east of Cape Agulhus, and then only with a sample from Cape Recife. The allele frequencies for the total Tasmanian farm sample and each of the four tanks were compared with those from the combination of seven wild samples, and then with a sample from Trumpeter Corner (east coast Tasmania) only.

3. Results 3.1. H. midae Across the two farmed samples 13 alleles were detected at locus CmrHr 2.15 (11 in western farm, seven in eastern farm), two at CmrHr 2.23 and 11 at CmrHr 2.29 (ten in western farm, five in eastern farm). These numbers compare with 19 at locus CmrHr 2.15 (18 west of Cape Agulhas, 11 east), two at CmrHr 2.23 and 17 at CmrHr 2.29 (17 west of Cape Agulhas, 10 east) in the wild samples respectively (Table 1). Two previously unrecorded alleles were identified at locus CmrHr 2.15, one in each of the farmed samples. Both new alleles were high repeat alleles at frequencies of 0.051 and 0.010. A single allele

114

B. Evans et al. / Aquaculture 233 (2004) 109–127

Table 1 Allele frequencies for three microsatellite alleles in wild and farmed samples of H. midae in South Africa Alleles (bp)

Wild west (108)

West Farm (50)

Wild East (127)

East Farm (52)

CmrHr2.15 243 249 251 253 257 261 263 265 267 269 271 273 275 276 277 279 281 283 293 303 309

0.014 0.016 – 0.010 0.010 0.034 0.016 0.031 0.328 0.029 0.005 0.018 0.024 0.006 0.379 0.019 0.046 0.010 0.005 – –

– – – 0.061 0.010 0.031 0.010 – 0.306 0.020 – – 0.020 – 0.398 0.082 0.010 – – – 0.051

0.008 0.007 0.004 0.015 – – – – 0.158 – – 0.016 0.008 – 0.628 0.066 0.068 0.024 – – –

– – – 0.010 – – – – 0.354 – – – 0.031 – 0.531 – 0.042 0.021 – 0.010 –

CmrHr2.23 243 253

0.895 0.105

0.898 0.102

0.857 0.143

0.830 0.170

CmrHr2.29 426 428 430 442 444 446 448 450 452 454 456 458 460 462 464 466 468

0.251 0.005 0.005 0.005 0.005 0.005 0.076 0.005 0.009 0.010 0.009 0.010 0.014 0.514 0.015 0.009 0.083

0.300 – – – – – 0.070 0.030 0.010 0.050 – 0.030 – 0.440 0.020 0.010 0.020

0.362 0.017 – – – – 0.035 0.004 – – – 0.046 0.027 0.392 0.004 0.007 0.105

0.610 – – – – – 0.040 – – – 0.010 – – 0.310 – – 0.030

Wild samples are combined frequencies from three collections to the west and three to the east of Cape Agulhus. Number of individuals in parentheses. – indicates allele not observed.

B. Evans et al. / Aquaculture 233 (2004) 109–127

115

at CmrHr 2.29 was recorded in the east farm sample (frequency of 0.010) that was not observed in the east coast wild samples. Common alleles in the wild remained common in the farm samples. All loci were polymorphic in each of the farmed samples examined. Numbers of alleles per locus can provide an accurate measure of genetic variation when the samples being compared are similar in sample size. To that end, each of the farmed samples was compared to a single wild sample from the area of broodstock collection, containing similar sample sizes (Table 2). A loss of eight of twenty alleles (40%) was observed in the eastern farm sample from that in the wild sample from Cape Recife. All lost alleles were at a frequency of less than 0.07 in the wild sample. A loss of 12 of 32 alleles (37.5%) was observed in the western farm sample when compared to the Kleinmond sample. All lost alleles were present at a frequency of less than 0.06 in the wild sample. Observed heterozygosities per locus were similar in all samples with the exception of locus CmrHr 2.29 in the western farm sample which was higher than all wild samples at Ho = 0.800. Table 2 H. midae genetic diversity estimates in samples from two South African hatcheries and their purported broodstock areas Sample Kleinmond

West Farm

Cape Recife

East Farm

CmrHr 2.15 n Nallele Ho He P D n Nallele Ho He P D n Nallele Ho He P D n Nallele Ho He P D

48 13 0.396 0.781 < 0.001** 0.493 49 11 0.408 0.767 < 0.001** 0.468 51 10 0.510 0.641 0.004** 0.204 48 7 0.417 0.596 < 0.001** 0.300

CmrHr 2.23 49 2 0.245 0.217 1.000 0.129 49 2 0.204 0.239 1.000 0.146 51 2 0.275 0.267 1.000 0.030 50 2 0.340 0.393 0.323 0.135

CmrHr 2.29 49 17 0.612 0.702 0.280 0.128 50 10 0.800 0.712 0.703 0.124 51 8 0.627 0.712 0.135 0.119 50 5 0.520 0.535 0.012** 0.028

All loci 48.7 32 0.418 0.567 < 0.001** 0.164 49.3 23 0.471 0.573 < 0.001** 0.164 51 20 0.396 0.490 < 0.025** 0.210 49.3 14 0.426 0.508 < 0.001** 0.154

n, sample size; Nallele, number of alleles; Ho, observed heterozygosity; He, expected heterozygosity; D, Selander’s index of heterozygote defficiency, negative values indicates an excess of homozygotes. P, probability of deviation from Hardy – Weinberg equilibrium. **Significant departure from Hardy – Weinberg equilibrium after sequential Bonferroni correction for multiple tests across loci. All loci provides mean values, with the exception of Nallele which is the sum of alleles across loci. All loci P value is calculated by combining probabilities across loci, and significance determined by comparison to critical values of chi-squared in Sokal and Rohlf (1981).

116

B. Evans et al. / Aquaculture 233 (2004) 109–127

This compares to values of 0.612 and 0.627 at this locus in the Kleinmond and Cape Recife samples (Table 2). Linkage disequilibrium was assessed and no significant departure from equilibrium levels was detected in any sample. Genotype proportions in the farmed samples were tested for goodness-of-fit to Hardy– Weinberg expectations at each locus (Table 2). Three of the six tests differed significantly from the Hardy– Weinberg expectations after sequential Bonferroni correction. Those tests that showed significant deviation from Hardy – Weinberg expectations were for CmrHr 2.15 in both farmed samples and CmrHr 2.29 in the eastern farm sample. All significant deviations from Hardy –Weinberg expected heterozygosities were due to an excess of homozygotes. Negative values of D (Selander, 1970) were obtained for five of the six tests, indicating an overall deficit of heterozygotes (Table 2). Similar observations were recorded for the wild samples. Sample-pairwise FST values were calculated between the east and west farm samples and pooled wild samples from either side of Cape Agulhas. Significant differentiation was observed between all pairs ( P < 0.001) with the exception of the pooled western Cape sample and the western farm sample. 3.2. H. rubra At the highly polymorphic locus CmrHr 2.30 (60 alleles in wild samples) no new alleles were observed in the farm samples (Table 3). However, three previously unrecorded alleles (all in Tank 2) were observed at the other highly polymorphic locus RUBCA1 (43 alleles in wild samples), and one at each of the other three loci. In tank 2, the new 124 bp allele at RUBCA1 was observed at a frequency of 0.349, making it the most common allele at this locus. In three of the four tanks, the most common allele observed at RUBCA1 was at a higher frequency than it occurred in wild samples. More common alleles were similar in farmed and wild samples for CmrHr 1.14, CmrHr 1.24 and CmrHr 2.14. At locus CmrHr 2.30 each of the four tanks had a different more common allele (frequency = 0.227, 0.383, 0.262, 0.172 respectively), which in each case was at a low frequency in the wild samples (frequency = 0.086, 0.100, 0.040, 0.051, respectively). All loci, with the exception of CmrHr 1.24 were polymorphic in all samples. Only the 222 bp allele was present at this locus in the tank 2 sample, this allele is at a frequency of greater than 0.70 in all other samples. Observed heterozygosity per locus ranged from 0.000 at that monomorphic locus, to 0.984 at CmrHr 2.30 in tank 1 (Table 4). When the tank samples were pooled to provide a single farm sample for comparison, the average observed heterozygosity was 0.574, compared to a value of 0.578 in the pooled wild sample. To ensure that sample sizes were similar for allele number comparisons, each farm tank (n = 64) was compared to the wild sample from Trumpeter Corner (east coast Tasmania) (n = 61). No significant differentiation had been found between samples from around Tasmania (Elliott et al., 2003 unpublished), therefore this sample was representative of the natural population from which the broodstock had been collected. The total number of alleles observed across five loci in this single wild sample was 83 (Table 4). Total number of alleles declined in each of the four tank samples. The largest loss of alleles (61%)

B. Evans et al. / Aquaculture 233 (2004) 109–127

117

occurred for the tank 2 sample (Nallele = 32), with the smallest loss of alleles (35%) being for the tank 4 sample (Nallele = 54). Because all tanks were representative samples of a single year class in one hatchery, they were pooled to create a single farm sample (n = 256) that showed a 35% reduction in alleles from the similar sized grouping of wild samples for three random sites (n = 260). All alleles lost in the farm samples were present at frequencies of less than 0.09 in all wild samples. Linkage disequilibrium was assessed and no significant departure from equilibrium levels was detected in any sample. Genotype proportions in all farmed samples were tested for goodness-of-fit to Hardy – Weinberg expectations at each locus (Table 4). Ten of the twenty tests differed significantly from the Hardy– Weinberg expectations after sequential Bonferroni correction. All farm samples showed significant departures from the expected frequencies at the RUBCA1 and CmrHr 2.30 loci, while all farm samples were in Hardy – Weinberg equilibrium at the CmrHr 1.14 and CmrHr 1.24 loci. The tank 2 and tank 4 samples both differed significantly from Hardy– Weinberg expected values at CmrHr 2.14. All significant deviations from Hardy – Weinberg expected heterozygosities in the wild samples were due to an excess of homozygotes (Table 4). In the farmed samples however, five of the ten significant values represent an excess of heterozygotes. The AMOVA method of Excoffier (1993) was used to provide a multi-locus estimate of UST, an analogue of FST, of 0.044 ( P < 0.001) across all samples. Population pairwise FST values indicate that the differentiation is derived from within the tank samples, with all tank samples significantly different ( P < 0.001) to all wild samples and all other tank samples (data not shown). Tank and wild samples were then grouped separately and compared in a second AMOVA. The FCT value, a measure of differentiation attributable to differences between the two groups was 0.013 ( P < 0.001), and the FSC value, a measure of the differentiation attributable to differences between the samples within those groups was 0.038 ( P < 0.001). Because the intention of this research was to assess genetic variation of hatchery reared stocks for ongoing production and potential re-stocking exercises, we were interested in the variation present across an entire year class, and not particularly within each individual spawning event. For this reason, tank samples were pooled as a single sample and the pairwise FST test was run again. This approach offers a more realistic assessment of the decline in genetic diversity within farmed stocks. The combined farm sample was significantly different to each wild sample ( P < 0.001). 3.3. Relatedness and effective number of breeders Mean relatedness of all wild samples (H. midae and H. rubra) approximated to zero ( 0.03 + 0.01 to 0.13 + 0.11) (Fig. 1). Mean relatedness of South African cultured samples also approximated to zero ( 0.04 + 0.02 for west farm 0.16 + 0.13 for east farm), whilst values for cultured H. rubra samples from each of the four tanks, and the combined farm sample were significantly greater than zero ( P < 0.025 Tank 4, and combined; P < 0.0005 Tanks 1– 3) (0.16 + 0.07 to 0.44 + 0.09). The effective number of H. midae breeders contributing to the west farm stock was estimated at 75.3 (SD F 57.6) based on the allele-frequency variance between the combined west coast samples and the farm sample. This estimate is reduced and with

118

B. Evans et al. / Aquaculture 233 (2004) 109–127

Table 3 Allele frequencies at five microsatellite loci in farmed and wild samples of H. rubra Allele (bp)

Wild (603)

Tank 1 (64)

Tank 2 (64)

Tank 3 (64)

Tank 4 (64)

Farm (256)

rubca1 110 114 116 118 120 122 124 126 128 130 132 134 136 138 140 142 144 146 148 150 152 154 156 158 160 162 164 166 168 170 172 174 176 178 180 182 184 186 188 190 192 194 196 198 202 208

0.018 0.001 – 0.267 – 0.013 – 0.015 0.009 0.001 0.002 0.016 0.004 0.012 0.007 0.029 0.050 0.009 0.032 0.015 0.051 0.014 0.037 0.083 0.049 0.038 0.057 0.025 0.035 0.017 0.011 0.020 0.014 0.009 0.006 0.002 0.003 0.001 0.003 0.004 0.008 0.006 0.002 0.001 0.001 0.001

– – – 0.117 – 0.156 – – – 0.023 – – – – – – – – 0.023 – 0.008 – – – 0.008 0.016 0.016 – – – – – 0.203 0.242 0.188 – – – – – – – – – – –

– 0.024 0.008 0.143 0.032 0.008 0.349 0.016 – – – – – – – – 0.008 – – 0.008 – 0.016 0.167 0.024 0.167 0.032 – – – – – – – – – – – – – – – – – – – –

– – – 0.127 – – – – – – – 0.024 – – 0.206 0.016 0.016 – 0.008 – 0.008 – 0.032 0.405 0.008 0.016 – – – 0.135 – – – – – – – – – – – – – – – –

0.031 – – 0.164 – 0.063 – – – – – 0.094 – 0.070 – 0.016 0.094 – 0.055 – 0.023 – – 0.055 0.047 0.031 0.070 – – 0.016 0.055 – 0.047 – – – 0.008 – – – – – 0.063 – – –

0.008 0.006 0.002 0.138 0.008 0.057 0.087 0.004 – 0.006 – 0.030 – 0.018 0.051 0.008 0.030 – 0.022 0.002 0.010 0.004 0.049 0.120 0.057 0.024 0.022 – – 0.037 0.014 – 0.063 0.061 0.047 – 0.002 – – – – – 0.016 – – –

CmrHr1.14 251 259

0.072 0.799

– 0.961

0.008 0.547

0.195 0.773

0.070 0.906

0.068 0.797

B. Evans et al. / Aquaculture 233 (2004) 109–127

119

Table 3 (continued) Allele (bp)

Wild (603)

Tank 1 (64)

Tank 2 (64)

Tank 3 (64)

Tank 4 (64)

Farm (256)

CmrHr1.14 261 263 265 267 269 271 275 277 283 289 291 293

0.091 0.006 0.002 0.014 0.005 0.004 0.002 0.002 0.001 0.002 – 0.001

0.039 – – – – – – – – – – –

0.438 – – – – – – – – – 0.008 –

0.031 – – – – – – – – – – –

0.008 – – 0.008 – – – – – 0.008 – –

0.129 – – 0.002 – – – – – 0.002 0.002 –

CmrHr1.24 212 216 220 222 224 226 228 230 236

0.001 0.012 0.002 0.816 0.108 0.042 0.017 – 0.003

– 0.008 – 0.836 0.016 – 0.117 0.023 –

– – – 1.000 – – – – –

– – – 0.984 0.008 – 0.008 – –

– – – 0.883 0.063 0.047 0.008 – –

– 0.002 – 0.926 0.021 0.012 0.033 0.006 –

CmrHr2.14 200 208 212 216 220 224 228 232 236 240 252

0.002 0.008 0.036 0.099 0.011 0.472 0.201 0.019 0.150 – 0.002

– – – 0.039 – 0.563 0.367 0.008 0.023 – –

– – – 0.211 0.008 0.531 0.070 0.008 0.172 – –

– – 0.024 0.056 0.008 0.603 0.294 0.016 – – –

– – 0.024 0.119 – 0.365 0.254 – 0.230 0.008 –

– – 0.012 0.106 0.004 0.516 0.246 0.008 0.106 0.002 –

CmrHr2.30 261 263 267 269 271 273 275 277 281 283 285

0.003 0.001 0.001 0.005 0.001 0.001 0.004 0.010 0.004 0.021 0.016

– – – – – – – – – – –

– – – – – – – – – – 0.047

– – – – – – – – – – –

– – – – – – – – – – –

– – – – – – – – – – 0.012

(continued on next page)

120

B. Evans et al. / Aquaculture 233 (2004) 109–127

Table 3 (continued) Allele (bp)

Wild (603)

Tank 1 (64)

Tank 2 (64)

Tank 3 (64)

Tank 4 (64)

Farm (256)

CmrHr2.30 287 289 291 293 295 297 299 301 303 305 307 309 311 313 315 317 319 321 323 325 327 329 331 333 335 337 341 343 345 347 349 351 353 355 357 359 361 363 365 367 369 371 373 375 377 379 381 385 389

0.005 0.031 0.046 0.037 0.047 0.060 0.053 0.071 0.046 0.062 0.035 0.029 0.026 0.029 0.024 0.027 0.028 0.008 0.023 0.014 0.019 0.022 0.002 0.006 0.005 0.014 0.009 0.011 0.008 0.006 0.003 0.010 0.009 0.006 0.009 0.006 0.009 0.009 0.009 0.008 0.006 0.006 0.009 0.007 0.006 0.008 0.002 0.006 0.002

– – – 0.039 – 0.227 0.055 – – 0.008 0.016 0.141 0.031 0.016 – 0.008 – – – – – – – 0.211 – 0.008 – 0.016 – – – – 0.055 0.008 – – – – – 0.008 – – – – – – 0.156 – –

– – 0.047 – – 0.008 – – – 0.383 – 0.016 – 0.258 – – – – – – – – – – – 0.234 – – – – – – – – 0.008 – – – – – – – – – – – – – –

– 0.167 – – 0.024 0.056 0.016 0.008 – 0.071 0.016 0.008 0.159 0.016 0.008 – – 0.008 – – – 0.262 0.159 – – 0.008 – – – – – – – – 0.008 – – – – 0.008 – – – – – – – – –

– 0.008 – – 0.070 0.008 0.016 0.063 0.031 0.063 0.023 0.164 0.031 – – 0.172 – 0.086 – – 0.016 0.055 0.023 – 0.047 – – 0.008 – – 0.031 – 0.008 0.008 0.023 – – – – 0.047 – – – – – – – – –

– 0.043 0.012 0.010 0.024 0.075 0.022 0.018 0.008 0.131 0.014 0.082 0.055 0.073 0.002 0.045 – 0.024 – – 0.004 0.078 0.045 0.053 0.012 0.063 – 0.006 – – 0.008 – 0.016 0.004 0.010 – – – – 0.016 – – – – – – 0.039 – –

Number of individuals in parentheses. Wild sample is combination of frequencies from seven sites. Farm is the combination of the four tank samples. – , indicates allele not observed.

B. Evans et al. / Aquaculture 233 (2004) 109–127

121

Table 4 H. rubra genetic diversity estimates in samples from four hatchery tanks and one wild population Population Trumpeter Corner

Tank 1

Tank 2

Tank 3

Tank 4

Farm sample

Wild sample

n Nallele Ho He P D n Nallele Ho He P D n Nallele Ho He P D n Nallele Ho He P D n Nallele Ho He P D n Nallele Ho He P D n Nallele Ho He P D

rubCA1

CmrHr 1.14

CmrHr 1.24

CmrHr 2.14

CmrHr 2.30

All

61 32 0.918 0.915 0.241 0.003 64 11 0.891 0.832 < 0.001** 0.071 63 14 0.714 0.820 < 0.001** 0.129 63 12 0.698 0.763 < 0.001** 0.085 64 18 0.906 0.930 < 0.001** 0.026 254 30 0.803 0.935 < 0.001** 0.141 583 43 0.842 0.902 0.146 0.067

61 5 0.328 0.386 0.145 0.150 64 2 0.047 0.091 0.077 0.487 64 4 0.500 0.551 1.000 0.093 64 3 0.453 0.389 0.132 0.165 64 5 0.141 0.189 0.090 0.254 256 6 0.285 0.347 < 0.001** 0.179 591 13 0.320 0.349 < 0.001** 0.083

61 5 0.377 0.411 0.729 0.083 64 5 0.328 0.289 0.838 0.135 64 1 – – – – 64 3 0.031 0.062 1.000 0.500 64 4 0.219 0.217 0.164 0.009 256 6 0.145 0.141 0.802 0.028 596 8 0.315 0.321 0.868 0.019

61 7 0.623 0.708 0.548 0.120 64 5 0.688 0.551 0.204 0.249 64 6 0.922 0.644 < 0.001** 0.432 63 6 0.587 0.571 0.267 0.028 63 6 0.619 0.744 0.003** 0.168 254 8 0.705 0.653 0.010** 0.080 542 10 0.653 0.703 < 0.001** 0.071

60 34 0.850 0.955 0.077 0.110 64 14 0.984 0.857 < 0.001** 0.149 64 7 0.969 0.733 < 0.001** 0.322 63 16 0.968 0.851 < 0.001** 0.137 64 21 0.797 0.919 < 0.001** 0.133 255 30 0.929 0.940 < 0.001** 0.011 572 60 0.760 0.967 < 0.001** 0.214

60.8 83 0.619 0.675 >0.100 0.083 64.0 37 0.588 0.524 < 0.001** 0.121 63.8 32 0.776 0.687 < 0.001** 0.130 63.4 40 0.547 0.527 < 0.001** 0.038 63.8 54 0.536 0.600 < 0.001** 0.106 255.0 80 0.574 0.603 < 0.001** 0.049 576.8 134 0.578 0.648 < 0.001** 0.109

N, sample size; Nallele, number of alleles; Ho, observed heterozygosity; He, expected heterozygosity; P, probability of deviation from Hardy – Weinberg equilibrium. **Significant departure from Hardy – Weinberg expected equilibrium after sequential Bonferroni correction for multiple tests across loci. All loci provides mean values, with the exception of Nallele which is the sum of alleles across loci. All loci P value is calculated by combining probabilities across loci, and significance determined by comparison to critical values of chi-squared in Sokal and Rohlf (1981). Wild samples refers to all seven Tasmanian wild samples.

122

B. Evans et al. / Aquaculture 233 (2004) 109–127

Fig. 1. Mean genetic relatedness and standard errors of wild and cultured abalone from Australia (light bars) and South Africa(dark bars).

less variability (43.5 F 29.8) if only the Kleinmond sample was used in the comparison. In both instances the observed loss of alleles from the three loci in the farm sample (15 when compared with total wild sample and 12 compared with Kleinmond) was significantly ( P < 0.001) greater than that expected (2.6 and 2.5, respectively) due to genetic drift based on such estimates of effective numbers of breeders. For the east farm stock of H. midae the effective number of breeders was estimated at only 18.5 ( F 8.4) when compared with the total east coast wild sample and 16.8 ( F 9.0) compared with only the Cape Recife sample. The observed loss of alleles in the farm sample (11 and 8) was again significantly greater ( P = 0.027 and P = 0.021, respectively) than expected (5.9 and 3.5). For the combined farm sample of H. rubra the effective number of breeders was estimated at 27.2 ( F 3.8) when allele-frequency variances were compared with the total wild sample, and 22.4 ( F 4.7) when compared to only the Trumpeter Corner sample. In both instances the observed loss of alleles (60 and 20) from the five loci was not significantly greater ( P = 0.095 and P = 0.544) than expected (50.2 and 18.2) due to genetic drift with such estimates of effective number of breeders. The estimated effective number of H. rubra breeders contributing to tank 4 (29.9 F 5.6 using all wild samples and

B. Evans et al. / Aquaculture 233 (2004) 109–127

123

19.1 F 4.8 using only Trumpeter corner) was about double that for the other three tanks (12.6 F 1.9 to 15.8 F 2.5 and 8.1 F 1.6 to 9.7 F 2.0, respectively), and this tank had the lowest mean relatedness (Fig. 1). The observed loss of alleles in individual tanks (80 –105 compared with total wild sample and 33 –56 compared with Trumpeter Corner only) in all instances was significant.

4. Discussion The five and three microsatellite loci used to investigate genetic variation in H. rubra and H. midae respectively reveal genetic differences between wild and hatchery stocks in both species. This is evident as a loss of rare alleles across most loci in all hatchery samples when compared to wild stocks (35 –62%), but not as a decline in overall heterozygosity. Tasmanian hatchery samples were genetically differentiated from Tasmanian wild samples, whilst the genetic structure of South African farm samples resembled wild samples from locations within the two areas of broodstock collection. The significance of the loss of alleles varies between the three farm examples in this study. Based on the estimated size of the effective hatchery broodstock, the loss of alleles in the two farm stocks of H. midae was significantly greater than expected due to genetic drift, while the loss for the H. rubra farm stock was not significant. The sample size was greater for H. rubra which may have some bearing on this result, however the significance of the loss in H. midae was independent of the wild sample used in the allele-frequency variance comparison. The H. rubra farm sample was a combination of four multiple tanks, and if each tank was examined independently the loss of alleles was significant. Therefore the significant results for H.midae may be a reflection of the sampling method, although the estimates of broodstock size especially for the west farm are high compared to both other farms. The loss of rare alleles from hatchery stocks has been reported as a more meaningful measure of genetic variation than heterozygosity. This is because heterozygosity is insensitive to the substantial genetic changes that may occur in cultivated aquaculture stocks within the first generations of culture (Hedgecock and Sly, 1990). In fact, Vuorinen (1984) states that the total extinction of any allele can be considered more harmful than a reduction in overall heterozygosity. The loss of alleles from all of the first generation hatchery stocks examined in this study must therefore be considered as a significant genetic alteration of cultured stocks. The loss of these alleles from the first generation stocks reduces the genetic variation available for future breeding programs using hatchery-reared abalone as broodstock and in re-seeding/stock enhancement programs. Whilst there is a minimum of 35% reduction in allele numbers in all farm samples in this study, this apparent decline in diversity of farm stocks is most likely exaggerated from real levels by the sampling methods employed. Each farm sample was collected from a single tank, representing only a fraction of the total number of tanks at each farm. By pooling allele frequencies from all cohorts on a farm, we would expect to see a more realistic estimation of genetic variation within the farm stock.

124

B. Evans et al. / Aquaculture 233 (2004) 109–127

Observed heterozygosity values of farmed and wild samples were similar over all loci in both species, with no apparent loss of heterozygosity in any of the farm samples measured. The loss of rare alleles without any noticeable reduction in heterozygosity can be an indication of a short-term population bottleneck (Nei et al., 1975; Allendorf, 1986) as would be the case during the foundation of farmed strains. Mgaya et al. (1995) revealed no loss in heterozygosity at three allozyme loci in either first or third generation cultured European abalone, Haliotis tuberculata, when compared to wild samples. They did however reveal the loss of one allele in the first generation hatchery sample, and two alleles in the third generation sample when each was compared to the wild sample (9 alleles). Smith and Conroy (1992) also noted a loss of rare alleles at allozyme loci in a first generation hatchery sample of Haliotis iris, which was associated with a reduction in heterozygosity at both allozyme loci examined. The highly polymorphic nature of microsatellites in abalone (Nallele locus-1 in H. rubra = 26.8) compared with allozymes (Nallele locus-1 in H. rubra = 6.1, Brown, 1991) make them highly sensitive for measuring change in genetic variation between hatchery and wild samples. Pairwise FST values support the association of farmed H. midae samples with the wild samples in the areas from which broodstock were collected. The western farm sample was shown to be most similar to a wild sample from Dassen Island, a common region for broodstock collection in the western Cape, whilst the eastern farm sample was significantly differentiated ( P < 0.001) from the Cape Recife sample, the stated source of broodstock collection, but very similar to a sample from the St. Francis area ( P = 0.541). No significant genetic differentiation was observed between wild samples of H. rubra from around Tasmania, whilst all tank samples, and a combined Tasmanian farm sample were genetically differentiated from all wild samples. Similar differentiation at allozyme loci has been reported between first generation hatchery and wild stocks of H. iris (Smith and Conroy, 1992), and multiple generation hatchery and wild stocks of brown trout, Salmo trutta (Ryman and Stahl, 1980). This study demonstrates that there is a reduction in genetic variation in the first generation Tasmanian and South African abalone hatchery stocks examined when compared to appropriate wild samples. As microsatellite markers are believed to be neutral, the loss of microsatellite alleles from the hatchery populations is in itself not a problem for the industry. The concern is that declines in variation at neutral markers such as these may be indicative of a loss of variation at coding regions. Minimizing this reduction is important to aquaculture, as this is the source of variation in important commercial traits such as growth rate and disease resistance (Vuorinen, 1984). When that variation is lost in first generation hatchery stocks, it is lost to all subsequent generations within a closed breeding program, and may therefore limit the genetic improvement available within that stock. Of perhaps more concern to industry is the change in frequency of some alleles, leading to change in the most common alleles observed. In tank 2 of the Tasmanian farm samples the RUBCA1 124 bp allele is recorded at a frequency of 0.349, despite not being recorded in the wild samples (603 individuals). The extent of local adaptation in abalone is not well understood, and therefore any major change in allele frequencies at neutral markers that is mirrored in coding regions could impact on the fitness of the species (Conover, 1998) in future breeding programs for aquaculture and restocking.

B. Evans et al. / Aquaculture 233 (2004) 109–127

125

Mean genetic relatedness is significantly higher in the four Tasmanian abalone tanks than that estimated for each wild sample, whilst the South African cultured samples are not significantly different to zero. These findings reveal that such husbandry practices as employed in the particular Tasmanian hatchery at the time of sampling may result in a high proportion of related individuals being produced for growout. In addition, if similar practices are followed for each spawning run (actual numbers of broodstock are unknown however) within the sampled farm, then we see a large variance in spawning success of assembled broodstock, differential fertilisation rates, larval or juvenile survival or settlement within each cohort. Sekino et al. (2003) recently published similar findings for a Japanese flounder species in which they utilised pedigree data to identify how few of the assembled broodstock had actually contributed to the progeny. This large discrepancy between the number of available breeders and those contributing to the composition of the progeny needs to be addressed for commercial production, but particularly for abalone culture for re-stocking programs. The situation in the South African west farm appears to be in a better state than that in the east farm and Tasmanian farm. However, the tank selection of the South African farmed samples is not known. Each farm supplied a single group of animals from 1-year class. If these animals were drawn from the complete range of spawning runs for the year, then we could expect to see very little relatedness. The Tasmanian samples were however each drawn from separate spawning runs, leading to a more informative picture of broodstock contribution. There is however, collaboration between the estimates of Nk and relatedness that supports the assumption of fewer breeders and more relatedness within Tasmanian tanks 1– 3 compared to tank 4.

5. Conclusion This study demonstrates the utility of highly polymorphic microsatellite markers as a means to compare genetic variation between wild and hatchery-reared abalone. They reveal a decline in the number of alleles from wild to farm samples in both H. rubra and H. midae. There is a demonstrated change in allele frequency from the Tasmanian wild samples to the hatchery samples, resulting in more common alleles becoming rare, and rare alleles becoming the most common in some cohorts. With the potential for local adaptation of abalone stocks it is possible that such large changes in microsatellite allele frequency may be indicative of similar changes in coding regions, and therefore could influence the fitness of hatchery stocks for continued culture or reseeding efforts. There is also a significant increase in the level of genetic relatedness in the particular Tasmanian farmed cohorts, and high levels of variation between spawning groups. The estimated number of effective breeders in the Tasmanian and eastern (SA) farms is low, whilst the estimate is much higher in the western farm. No change was observed in overall heterozygosity in either of the species examined. The data highlights the need for genetic monitoring of aquaculture hatchery systems to ensure that sufficient numbers of known pedigree broodstock are utilised in each generation, and that a high proportion of those individuals actually contribute to the progeny. These safeguards will then ensure that potentially valuable genetic variation is not lost from the system.

126

B. Evans et al. / Aquaculture 233 (2004) 109–127

Acknowledgements We thank the following people for helpful comments on drafts of this manuscript; Odette Ison, Bob Ward and three anonymous reviewers. This work was supported by a grant from Aquaculture CRC, Australia, and FRDC grant No. 1999/164. References Allendorf, F.W., 1986. Genetic drift and the loss of alleles versus heterozygosity. Zoological Biology 5, 181 – 190. Allendorf, F.W., Ryman, N., 1988. Genetic management of hatchery stocks. In: Ryman, N., Utter, F.W. (Eds.), Population genetics and fishery management. University of Washington Press, Seattle (418 pp). Benzie, J.A.H., Williams, S.T., 1996. Limitations in the genetic variation of hatchery produced batches of the giant clam Tridacna gigas. Aquaculture 139, 225 – 241. Boudry, P., Collet, B., Cornette, F., Hervouet, V., Bonhomme, F., 2002. High variance in reproductive success of the Pacific oyster (Crassostrea gigas, Thunberg) revealed by microsatellite-based parentage analysis of multifactorial crosses. Aquaculture 204, 283 – 296. Brown, L.D., 1991. Genetic variation and population structure in the blacklip abalone, Haliotis rubra. Australian Journal of Marine and Freshwater Research 42, 77 – 90. Burton, R.S., Tegner, M.J., 2000. Enhancement of red abalone Haliotis rufescens stocks at San Miguel Island: reassessing a success story. Marine Ecology Progress Series 202, 303 – 308. Conover, D.O., 1998. Local adaptation in marine fishes: evidence and implications for stock enhancement. Bulletin of Marine Science 62, 477 – 493. Danzmann, R.G., Ferguson, M.M., Allendorf, F.W., 1989. Genetic variability and components of fitness in hatchery strains of rainbow trout. Journal of Fish Biology 35(A), 313 – 319. Elliott, N.G., Reilly, A., 2003. Likelihood of bottleneck event in the history of the Australian population of Atlantic salmon (Salmo salar L.). Aquaculture 215, 31 – 44. Evans, B., 2002. Molecular markers for abalone research. Unpublished PhD thesis, University of Tasmania, Hobart. 149 pp. Evans, B., White, R.W.G., Elliott, N.G., 2000. Characterisation of microsatellite loci in the Australian Blacklip abalone (Haliotis rubra, Leach). Molecular Ecology 9, 1183 – 1184. Excoffier, L., 1993. WINAMOVA (vers. 1.5): Analysis of Molecular Variance. Computer Program University of Geneva, Geneva. Gaffney, P.M., Powell, R.V., Hedgecock, D., Powers, D.A., Morris, G., Hereford, L., 1996. Genetic effects of artificial propagation: signals from wild and hatchery populations of red abalone in California. Aquaculture 143, 257 – 266. Grewe, P.M., Krueger, C.C., Aquadro, C.F., Bermingham, E., Kincaid, H.L., May, B., 1993. Mitochondrial DNA variation among lake trout (Salvelinus namaycush) strains stocked into Lake Ontario. Canadian Journal of Fisheries and Aquatic Science 50, 2397 – 2403. Hahn, K.O., 1989. Handbook of Culture of Abalone and Other Marine Gastropods. CRC Press, Boca Raton, Florida. 348 pp. Hedgecock, D., Sly, F., 1990. Genetic drift and effective sizes of hatchery-propagated stocks of the Pacific oyster Crassostrea gigas. Aquaculture 88, 21 – 38. Hedgecock, D., Chow, V., Waples, R.S., 1992. Effective population numbers of shellfish broodstocks estimated from temporal variance in allelic frequencies. Aquaculture 108, 215 – 232. Koehn, R.K., Diehl, W.J., Scott, T.M., 1988. The different contribution by individual enzymes of glycolysis and protein catabolism to the relationship between heterozygosity and growth rate in the coot clam Milinia lateralis. Genetics 118, 121 – 130. Mgaya, Y.D., Gosling, E.M., Mercer, J.P., Donlon, J., 1995. Genetic variation at three polymorphic loci in wild and hatchery stocks of the abalone Haliotis tuberculata Linnaeus. Aquaculture 136, 71 – 80. Nei, M., Maruyama, T., Chakraborty, R., 1975. The bottleneck effect and genetic variability in populations. Evolution 29, 1 – 10.

B. Evans et al. / Aquaculture 233 (2004) 109–127

127

Norris, A.T., Bradley, D.G., Cunningham, E.P., 1999. Microsatellite genetic variation between and within farmed and wild Atlantic salmon (Salmo salar) populations. Aquaculture 180, 247 – 264. Queller, D.C., Goodnight, K.F., 1989. Estimating relatedness using genetic markers. Evolution 43, 258 – 275. Raymond, M., Rousset, F., 1995. GENEPOP (Vers. 1.2): population genetic software for exact tests and ecumenicism. Journal of Heredity 86, 248 – 249. Ryman, N., Stahl, G., 1980. Genetic changes in hatchery stocks of brown trout (Salmo trutta). Canadian Journal of Fisheries and Aquatic Science 37, 82 – 87. Schneider, S., Kueffer, J.M., Roessli, D., Excoffier, L., 2000. ARLEQUIN (vers. 2.000): A Software for Population Genetic Data Analysis. Genetics and Biometry Laboratory, University of Geneva, Geneva. Selander, R.K., 1970. Behavior and genetic variation in natural populations. American Zoologist 10, 53 – 66. Sekino, M., Saitoh, K., Yamada, T., Kumagai, A., Hara, M., Yamashita, Y., 2003. Microsatellite-based pedigree tracing in a Japanese flounder Paralichthys olivaceus hatchery strain: implications for hatchery management related to stock enhancement program. Aquaculture 221, 255 – 263. Smith, P.J., Conroy, A.M., 1992. Loss of genetic variation in hatchery produced abalone, Haliotis iris. New Zealand Journal of Marine and Freshwater Research 26, 81 – 85. Sokal, R.R., Rohlf, F.J., 1981. Biometry, 2nd ed. WH Freeman and Company, New York. 859 pp. Sweijd, N.A., 1999. Molecular Markers and Abalone Seeding as Tools for the Conservation and Management of the South African Abalone (Perlemoen), Haliotis midae (Linn.) resource. Unpublished PhD thesis, University of Cape Town Cape Town. 265 pp. Verspoor, E., 1988. Reduced genetic variability in first-generation hatchery populations of Atlantic salmon (Salmo salar). Canadian Journal of Fisheries and Aquatic Science 45, 1686 – 1690. Villanueva, B., Verspoor, E., Visscher, P.M., 2002. Parental assignment in fish using microsatellite genetic markers with finite numbers of parents and offspring. Animal Genetics 33, 33 – 41. Vuorinen, J., 1984. Reduction of genetic variability in a hatchery stock of brown trout, Salmo trutta. Journal of Fish Biology 24, 339 – 348.