Genetic effects of supportive stockings on native pikeperch populations in boreal lakes - three cases, three different outcomes

Genetic effects of supportive stockings on native pikeperch populations in boreal lakes – three cases, three different outcomes MATTI SALMINEN 1 , MARJA-LIISA KOLJONEN 1 , MARJATTA S Ä IS Ä 2 and JUKKA RUUHIJ Ä RVI 3 1 Finnish Game and Fisheries Research Institute, Helsinki, Finland 2 Department of Agricultural Sciences, University of Helsinki, Helsinki, Finland 3 Finnish Game and Fisheries Research Institute, Evo, Finland Salminen, M., Koljonen, M.-L., S ä is ä , M. and Ruuhij ä rvi, J. 2011 . Genetic effects of supportive stockings on native pikeperch populations in boreal lakes – three cases, three different outcomes. – Hereditas 149 : 1–15. Lund, Sweden. eISSN 1601-5223. Received 27 May 2011. Accepted 20 September 2011. The genetic consequences and gene fl ow of pikeperch ( Sander lucioperca ) stocking were assessed in three boreal lakes based on admixture model analysis and comparison of the preand post-release patterns of genetic variability at 9 DNA microsatellite loci in the recipient populations. In two out of the three cases, the releases of fi sh from foreign populations caused signifi cant changes in the genetic structure of the recipient population. The largest changes were observed in Lake Ouluj ä rvi, where the post-release sample was almost identical to the released Lake Vanajanselk ä population, and about 90% of the catch was composed of the released population. The genetic composition of Lake Lohjanj ä rvi pikeperch also shifted markedly towards that of the released Lake Vanajanselk ä population, and about half of the later catch was of released Vanajanselk ä origin. In Lake Vanajanselk ä , in contrast, releases of pikeperch from lakes Painio and Averia had only a small impact on the genetic structure of the pikeperch population. These results indicate that the current stocking practices create an effective artifi cial gene fl ow that may strongly shape and reduce the genetic differentiation among the remaining native pikeperch populations. A common feature of all three cases was the lack of prior appraisal of the potential genetic and ecological risks in relation to the expected benefi ts of the release programmes. Matti Salminen, Finnish Game and Fisheries Research Institute, PO Box 2, FIN-00791 Helsinki, Finland. E-mail: matti.salminen@ rktl.fi The native distribution of pikeperch ( Sander lucioperca ) in northern Europe is assumed to be related to the Lake Ancylus (freshwater) stage of the present Baltic Sea (L Ö NNBERG 1899; LEHTONEN et al. 1996), which ca 9200 – 9000 BP (B J Ö RCK 1995) provided the species a distribution path to areas covered by the former lake, up to 100 – 150 m above the present water level of the Baltic Sea. Since the initial colonization, adaptive selection, gene fl ow and drift have shaped the patterns of genetic diversity within the species, resulting in a relatively high level of genetic variability (B J Ö RKLUND et al. 2007; S Ä IS Ä et al. 2010) comparable to that reported in Scandinavian whitefi sh ( Coregonus lavaretus, S Ä IS Ä et al. 2008), grayling ( Thymallus thymallus, K OSKINEN et al. 2000), perch ( Perca fl uviatilis , NESB Ø et al. 1999) and Atlantic salmon populations ( Salmo salar, S Ä IS Ä et al. 2005). Since the late 1800s, the distribution and genetic diversity of pikeperch has also been shaped by human-infl uenced gene fl ow through introductions and hatchery releases. In Denmark, the fi rst introduction took place in 1898, and since then pikeperch have been established in over 70 water bodies ( DAHL 1982). In Sweden, Norway and Finland the pikeperch is a native species, but has also been the most commonly introduced non-salmonid fi sh ( TAMMI et al. 2003). In Finland, the pikeperch is native in ca 650 lakes, but due to introductions, the present distribution includes ca 2300 lakes ( LAPPALAINEN and TAMMI 1999). The expansion of pikeperch has been exclusively regarded as a positive phenomenon, unlike in Turkey, the UK and Denmark, for instance, where the expansion has in some cases involved undesirable ecological effects ( CRIVELLI 1995; COWX et al. 1997; JEPSEN et al. 2000). Until the mid-1900s, introductions in Finland were mostly carried out using repeated transfers of adult fi sh or fertilized eggs ( HALME 1961, 1962), a method that seems to have been quite effective in transferring genetic material ( S Ä IS Ä et al. 2010). Some of the most vital naturally reproducing pikeperch populations in Finland stem from these early introductions, including two major source stocks of current hatchery production. Apart from the stocking pressure (no. of stocking events and individuals released; KOLAR and LODGE 2001) and the traits of the species, the success of the early introductions may relate to the availability of favourable habitats at altitudes above the Lake Ancylus water level, and to the fact that the source stocks were usually located close to the recipient lakes. In the Rh ô ne delta of France, the invasive success of pikeperch has similarly been explained by the traits of the species and the stocking pressure, maintaining a high level of genetic variability in the introduced populations ( POULET et al. 2008). A new era of pikeperch stocking was catalyzed by the collapse of several important pikeperch stocks during the Hereditas 149: 1–15 (2012) © 2012 The Authors. This is an Open Access article. DOI: 10.1111/j.1601-5223.2011.002230.x 2 M. Salminen et al. Hereditas 149 (2012) 1960s and early 1970s ( COLBY and LEHTONEN 1994). The management strategy devised in this situation was based on the stocking of young-of-the-year (YOY) pikeperch. More effi cient rearing methods were developed ( RUUHIJ Ä RVI and HYV Ä RINEN 1996), increasing the production of YOY pikeperch to ca 10 million year ⫺1 ( ANON . 2004). The production was mainly directed to new introductions and re-stocking projects aiming at re-establishing lost pikeperch populations. In recent years, the fi sh have mainly been used in various enhancement projects aimed at the mitigation of reproduction failures or the effects of overfi shing, or simply at improving fi shing possibilities. The popularity of enhancement programmes has raised questions about their sustainability. Despite the high costs of the programmes, their economic benefi ts have often not been properly evaluated. Stocking also has been a tempting solution in many cases, as it reduces the need for strict fi shing regulation. The main concern, however, is the potential adverse effect of releases of foreign genetic material on the genetic diversity and adaptability of the species as a whole. Due to insuffi cient control and planning, practically all stocking programmes in Finland have relied on three to four source stocks, all of them with a southern origin. The possibly small effective population size of the broodstocks used has also raised concerns, as it causes loss of diversity ( RYMAN and L AIKRE 1991). During the last two decades, practically all remaining indigenous freshwater pikeperch stocks are likely to have been subjected to the fl ow of foreign southern genes through hatcheries. The genetic effects of releases such as competition between native and introduced species or stocks, the replacement of native stocks, mixing of stocks and extinction of native stocks have been widely discussed ( COWX et al. 1997). In this study, we assessed the genetic consequences of pikeperch stockings for indigenous native populations in three boreal lakes in Finland: Lake Lohjanj ä rvi, Lake Vanajanselk ä and Lake Ouluj ä rvi. Genetic information was available from the three recipient pikeperch populations before releases, from the three released hatchery populations and also the admixed populations in each lake after the releases, allowing us to assess the proportion of new genetic material in the native populations after the releases. Finally, to address the overall sustainability of current stocking practices, we contrasted the observed genetic effects of the three stocking programmes with their expected socioeconomic benefi ts. MATERIAL AND METHODS Study lakes Lake Lohjanj ä rvi The indigenous pikeperch is recreationally and economically one of the most important fi sh species in Lake Lohjanj ä rvi (area 89 km 2 , mean depth 13 m), southern Finland (Fig. 1). In mail surveys, reported pikeperch catches from 1981 – 2002 varied from 4 – 21 tons year ⫺1 ( SALMINEN and RUUHIJ Ä RVI 2004). About 90% of the catches have been taken by gillnets, while trolling accounts for 10% of the total yield. The catches are mainly used for household needs, but some fi shermen also sell fi sh. L EHTONEN and MIINA (1988) reported high fi shing mortality and a low age at recruitment (4 – 6 years) for Lake Lohjanj ä rvi pikeperch. Growth over-fi shing was regarded as evident and recruitment over-fi shing possible. A larger gill-net mesh size (50 – 55 mm, bar length) and a larger minimum landing size (MLS; 40 – 42 cm) were suggested and in 1992 also implemented to increase the age at recruitment. Releases of YOY pikeperch were also expanded to increase and stabilize catches, and in the 1990s they reached the level of 100 000 individuals year ⫺1 (ca 11 ind. ha ⫺1 , Fig. 2). N 0100 km ESRI 1999 © Oulujärvi Lohjanjärvi Vanajanselkä Averia Painio Kivijärvi Kemijärvi FINLAND Fig. 1. The location of the three study lakes (blue dots; Lakes Lohjanj ä rvi, Vanajanselk ä and Ouluj ä rvi) and the three pikeperch populations additionally used in the releases (green dots; Lakes Averia, Painio and Kivij ä rvi). The pikeperch population of Lake Kemij ä rvi (yellow dot) was used as an outgroup in the genetic distance analyses. Hereditas 149 (2012) Genetic effects of supportive stockings on pikeperch 3 Lake Lohjanjärvi 0 20 000 40 000 60 000 80 000 100 000 120 000 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 B2: 1998-2002B1: 1991-1995 Lake Vanajanselkä 0 10 000 20 000 30 000 40 000 50 000 60 000 70 000 80 000 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 A: 1984-1986 B: 1998-2002 Lake Oulujärvi 0 100 000 200 000 300 000 400 000 500 000 600 000 700 000 800 000 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 Averia Vanajanselkä Painio A: 1936-1943, 1961-1968 B2: 2001B1: 1990-1992 A: 1979-1981 Fig. 2. Releases of YOY pikeperch according to the year-class (1979 – 2003) and source stock in Lake Lohjanj ä rvi, Lake Vanajanselk ä and Lake Ouluj ä rvi. Arrows indicate the year-classes of pre- (A) and post-stocking (B) samples (Table 1). Lake Vanajanselk ä Lake Vanajanselk ä is the largest lake (103 km 2 , mean depth 8 m) of the Vanajavesi watercourse in the River Kokem ä enjoki basin. The relatively strong and stable indigenous pikeperch population ( TOIVONEN et al. 1981) supports important pikeperch fi sheries and has also served as a source for early transfers of adult fi sh and eggs ( HALME 1961, 1962; RUUHIJ Ä RVI and SALMINEN 1992), and since the 1980s also for large-scale YOY production ( RUUHIJ Ä RVI and HYV Ä RINEN 1996). Pikeperch catches of recreational fi shermen have recently been ca 20 tons year ⫺1 . Seven professional fi shermen (in 2006, Pekka Korhonen pers. comm.) also fi sh pikeperch in the lake, their catch being 5 – 10 tons year ⫺1 . Gill-netting with mainly 45 mm nets accounts for ca 90% of the pikeperch catch, the rest being caught mainly by trollers ( KIVINEN 2009). Despite the good growth rate of Lake Vanajanselk ä pikeperch, the recommended rises in the minimum landing size (present 37 cm) and gill-net mesh size have not been implemented. Instead, enhancement releases have been carried out (Fig. 2), despite the good recruitment of the native population. Lake Ouluj ä rvi Until the late 1950s, Lake Ouluj ä rvi (area 928 km 2 , mean depth 7.6 m) pikeperch sustained a fl ourishing fi shery, with catches amounting to 100 – 150 tons year ⫺1 ( SUTELA and HYV Ä RINEN 2002). Thereafter, catches began to decline and the bottom, less than 100 kg year ⫺1 , was reached in the early 1980s. In the 1990s, catches began to recover and have recently been around 100 tons year ⫺1 . Pikeperch are mainly taken by recreational gill-netters and anglers, but also provide an important target for the 32 professional fi shermen operating in the lake (in 2006, Pekka Korhonen pers. comm.). Potential causes for the collapse of the pikeperch stock and catches in Lake Ouluj ä rvi, as in numerous other boreal lakes, were the declining temperatures in 1940 – 1970 combined with increasing exploitation caused by the shift from cotton to more effi cient nylon monofi lament gillnets ( COLBY and LEHTONEN 1994). Enhancement stocking was the main management action recommended in this situation ( SALOJ Ä RVI et al. 1981), and based on this recommendation releases were started in 1985, reaching the level of 500 000 individuals year ⫺1 in the 2000s (Fig. 2, ca 5 ind. ha ⫺1 ). The production and releases of hatchery pikeperch The production of YOY pikeperch relies on the annual capture of wild spawners from the spawning areas using trap-nets ( SALMINEN and RUUHIJ Ä RVI 1991; RUUHIJ Ä RVI and H YV Ä RINEN 1996). After hatching the larvae are stocked at 20 000 – 40 000 ind. ha ⫺1 into 1 – 2 ha dams producing natural food. The average harvest in late August – September, i.e. after 3 – 4 months of rearing, is about 10 000 YOY pikeperch ha ⫺1 , ranging in total length from 60 – 80 mm. As the background of YOY pikeperch has not been an issue among the managers, the production of larvae has concentrated on a limited number of dense source populations, all with a southern origin. Another factor potentially infl uencing the genetic consequences of releases is family size, i.e. the number of juveniles produced per female. Given the high fecundity in artifi cial rearing ( SALMINEN et al. 1992) and high survival during rearing, one large female may produce up to 200 000 juveniles, which may 4 M. Salminen et al. Hereditas 149 (2012) 493 (0.5 – 3.0 kg) adult fi sh from Lake Vanajanselk ä (Fig. 3, Table 1) ( SUTELA et al. 1995). Pikeperch samples The analysis of genetic impacts was based on admixture modelling and the comparison of genetic variability within the three recipient populations in Lakes Lohjanj ä rvi, Vanajanselk ä and Ouluj ä rvi, before and after stocking. From all three lakes, one pre-stocking sample and one or two (in Lake Ouluj ä rvi) post-release admixture samples were analyzed (Table 1). The three pikeperch populations additionally used in the releases (from lakes Averia, Painio and Kivij ä rvi) were also sampled for the DNA analyses. The pikeperch population of Lake Kemij ä rvi was used as an outgroup in the genetic distance analyses ( S Ä IS Ä et al. 2010). In the case of Lohjanj ä rvi it was possible to organize an additional test, independent of the genetic data, as the juveniles (from Lake Averia and Lake Vanajanselk ä ) released there in 1991 – 1995 were marked using hotbranding ( SAURA 1996; SALMINEN and RUUHIJ Ä RVI 2004), allowing the identifi cation of their stock of origin (Table 1, Fig. 2). Their stock-specifi c proportions were recorded among adult pikeperch sampled in 1994 – 2000. then constitute a large part of or even the whole year-class stocked in one or two larger or several smaller lakes. To prevent this, hatcheries have been advised to mix the offspring of females that spawn at approximately the same time, but it is not known to what extent this actually happens. Our three study lakes offer typical examples of the complex transfers of pikeperch from one water-body to another: 1) In Lake Lohjanj ä rvi, three foreign populations were used in the releases (Fig. 3, Table 1). Two of them (Lake Averia and Lake Painio) are genetically close to Lake Lohjanj ä rvi pikeperch ( S Ä IS Ä et al. 2010), as they have their roots in early (1930s) transfers from the same Lake Lohjanj ä rvi, while the third source population (Lake Vanajanselk ä ) is genetically more distant ( S Ä IS Ä et al. 2010). 2) In Lake Vanajanselk ä , both its own indigenous population, and the populations of Lake Averia and Lake Painio have been used in the releases (Fig. 3, Table 1). 3) In Lake Ouluj ä rvi, pikeperch from Lake Vanajanselk ä have been stocked, mostly via a special broodfi sh lake (Lake Kivij ä rvi) housing a pikeperch population established in the late 1980s by transferring Lake Lohjanjärvi Lake Vanajanselkä Lake Oulujärvi Lake Kivijärvi Lake Averia Lake Painio 1985–2003 1981–2003 1980s 1993–2003 1999–2001 1930s 2000 1930s 1993–1996 Averia (Lohjanjärvi) - S1/original (1984) - S2/from Lohjanjärvi (1994–2000) Painio (Lohjanjärvi) - S3/original (2006) Case Lohjanjärvi - S4/original (1982–1986) - S5/wild (1994–2000) - S6/mixed (2002–2003) Case Vanajanselkä - S7/original (1988–1991) - S8/from Lohjanjärvi (1994–2000) - S9/mixed (2001–2006) Case Oulujärvi - S11/original (1946/1974) - S12/mixed - S13/mixed (2002) Kivijärvi (Vanajanselkä) - S10/original (2008) Fig. 3. Genetic background and population transfer history of the sampled Finnish pikeperch populations. Dashed lines indicate transfers of adult fi sh, solid lines transfers of YOY. Sample numbers (S) refer to Table 1. Hereditas 149 (2012) Genetic effects of supportive stockings on pikeperch 5 between populations were assessed by the ttest for paired observations ( NEI 1987). Analysis of the differences between samples was based on allele frequency differences, using pairwise F st values ( WEIR and COCKERHAM 1984), which were estimated with FSTAT ver. 2.9.3 ( GOUDET 2001). Standard deviations and confi dence intervals were estimated through bootstrapping. Genetic distances between samples were calculated using Nei ’ s D A distances ( NEI et al . 1983). A phylogenetic tree was constructed using a neighbour joining (NJ) algorithm ( SAITOU and NEI 1987) with DISPAN software ( OTA 1993). Bootstrapping with 1000 replicates was used to test the statistical strength of the branches. To analyse the population mixtures in the recipient lakes, Bayesian clustering algorithms were used in the program STRUCTURE (ver. 2.2.3) ( PRITCHARD et al. 2000; F ALUSH et al. 2003; PRITCHARD and WEN 2004). The program assigns individual fi sh to one or more groups, with their relative frequency of predicted membership totalling 1.00. This allowed analyses of admixed populations with prior information from the source populations. In all cases, the options of using population information and applying the admixture model for the unknown catch sample were used. Moreover, alpha , describing the amount of population mixing, was allowed to vary in all cases and was different for each population. The total length of the runs was 150 000 iterations, with a burn-in of 100 000 iterations and the last 50 000 iterations being used for the estimates. DNA analysis and calculations DNA extraction and microsatellite laboratory analysis were conducted according to the description of S Ä IS Ä et al. (2010). Variation in the following nine microsatellite loci was determined: Pfl aL3, Pfl aL8 ( LECLERC et al. 2000), Svi4, Svi6, Svi18, Svi33 ( BORER et al. 1999), SviL7, SviL8 and SviL11 ( WIRTH et al. 1999). The number of alleles in samples was compared using a rarefaction-based allelic richness measure ( EL MOUSADIK and PETIT 1996; PETIT et al. 1998), which was calculated with FSTAT software ver. 2.9.3 ( GOUDET 2001). The program calculates allelic richness for the smallest number of individuals typed for any locus. Each locus was calculated separately with the same number over all populations, and the mean was calculated over loci. Population differentiation was analyzed with the GENEPOP (ver. 4.0) software package ( RAYMOND and ROUSSET 1995; ROUSSET 2008) with Markov chain parameters, 300 batches and 3000 iterations. The Bonferroni correction ( RICE 1989) was applied to correct for the number of tests in the H-W equilibrium. The potential occurrence of null alleles was also checked by GENEPOP. Indication of a null allele in locus Svi33 could be seen, but it was still included in the analysis, as indication of null alleles may result from Hardy-Weinberg deviations as well. The expected heterozygosity level in each sample was calculated using Popgene ver. 1.32 ( YEH and BOYLE 1997). Differences in the mean heterozygosities and allele richness ’ s Table 1. Analysed pikeperch samples. Sample number, genetic background of the population, sampling site (Fig. 1), sampling period (A ⫽ pre-stocking, B ⫽ post-stocking), sample size (n), catch year, year class and status of studied pikeperch samples. * Individuals for the sample were recognized from hot branding marks. Sample Genetic background Sampling lake Sampling period n Catch year Year class Status S1 Lohjanj ä rvi (broodstock) Averia – 63 1984 1979 – 1982 original/source S2 Lohjanj ä rvi Lohjanj ä rvi B1 59 1994–2000 1991 – 1993 Lake Averia origin * S3 Lohjanj ä rvi (broodstock) Painio – 74 2006 1985 – 2005 original/source S4 Lohjanj ä rvi Lohjanj ä rvi A 60 1982–1986 1979 – 1981 Original S5 Lohjanj ä rvi Lohjanj ä rvi B1 60 1994–2000 1991 – 1995 wild * S6 Lohjanj ä rvi and Vanajanselk ä Lohjanj ä rvi B2 200 2002–2003 1998 – 2002 current, mixed S7 Vanajanselk ä Vanajanselk ä A 60 1988–1991 1984 – 1986 original S8 Vanajanselk ä Lohjanj ä rvi B1 60 1994–2000 1993 – 1995 Vanajanselk ä origin * S9 Vanajanselk ä and Lohjanj ä rvi Vanajanselk ä B 208 2001–2006 1998 – 2002 current, mixed S10 Vanajanselk ä (broodstock) Kivij ä rvi – 31 2008 1992 – 1998 original/source S11 Ouluj ä rvi Ouluj ä rvi A 8, 15 1946, 1974 1936 – 1943, 1961 – 1968 original S12 Ouluj ä rvi and Vanajanselk ä Ouluj ä rvi B1 100 1990s 1990 – 1992 mixed S13 Ouluj ä rvi and Vanajanselk ä Ouluj ä rvi B2 100 2002 2001 current, mixed 6 M. Salminen et al. Hereditas 149 (2012) Fig. 4a – d. Estimated population structure of pikeperch from Bayesian STRUCTURE analysis for four catch mixtures. Each individual is represented by a thin vertical line, which is partitioned into K coloured segments that represent its estimated population group membership fractions. Black lines separate individuals from different numbered samples. Sample names and numbers are given separately for each case. ( a ) Lake Lohjanj ä rvi 2000, Population 1 ⫽ Averia, red (S1), 2 ⫽ Painio, green (S3), 3 ⫽ Vanajanselk ä , blue (S7), and 4 ⫽ original Lohjanj ä rvi, yellow (S4), 5 ⫽ catch mixture of Lake Lohjanj ä rvi in 2000 (S6). ( b ) Lake Vanajanselk ä 2000, (Population 1 ⫽ Averia, red (S1), 2 ⫽ Painio, green (S3) and 3 ⫽ original Vanajanselk ä , blue (S7), 4 ⫽ catch mixture from Lake Vanajanselk ä in 2000 (S9). ( c ) Lake Ouluj ä rvi 1990. Population 1 ⫽ Kivij ä rvi, red (S10), 2 ⫽ Ouluj ä rvi original, green (S11) and 3 ⫽ catch mixture from Lake Ouluj ä rvi in the 1990s (S12). ( d ) Lake Ouluj ä rvi 2000. Population 1 ⫽ Kivij ä rvi (S10), 2 ⫽ Ouluj ä rvi original (S11) and 3 ⫽ catch mixture from Lake Ouluj ä rvi in 2000 (S13). Numbers in brackets refer to Table 1, sample number. Convergence of all runs was checked. In all cases, the number of contributing populations, K, was known, which simplifi ed the analyses. Several runs were carried out for each admixture and the consistency of the runs was assessed. Changes in K were also tested with K values of one more or less than the known number of populations, and results with the greatest posterior probabilities are presented. Admixture analysis was also performed with maximum likelihood estimation by WANG (2003), included in the LEADMIX software. This is based on the principle that allele frequencies of the admixed populations should be Hereditas 149 (2012) Genetic effects of supportive stockings on pikeperch 7 heterozygosity (H e ⫽ 0.54) in the pre-stocking sample of Lake Ouluj ä rvi (S11; Table 2). The lowest allelic richness (3.6 alleles) was observed in the Lake Painio population (S3) and in the pre-stocking sample from Lake Lohjanj ä rvi (S4), and the lowest expected heterozygosity in the original Lake Averia population (S1; 0.41). The genotype distributions of the three native populations in lakes Lohjanj ä rvi, Ouluj ä rvi and Vanajanselk ä did not deviate from Hardy–Weinberg equilibrium after Bonferroni correction. The genetic diversity was in general relatively high in all cases both before and after stock transfers (Table 2), and no decrease in genetic diversity could be observed as a result of stocking. In all three cases, allelic richness was actually somewhat higher after releases than in the original pre-stocking population. In Lakes Lohjanj ä rvi and Vanajanselk ä , the mean heterozygosity was also higher in the contemporary than in the native population. The only statistically signifi cant difference in the mean heterozygosity was between the Lake Painio population (S3) and the contemporary admixed population of Lake Lohjanj ä rvi (S6), which had a higher diversity. Allelic richness was also statistically signifi cantly higher in the contemporary Lohjanj ä rvi sample (S6) than in the samples from Lake Averia (S1), Lake Painio (S3) and even in the sample from the native Lake Lohjanj ä rvi population (S4). Samples S2 and S8 from Lake Lohjanj ä rvi, both representing three consecutive year-classes of adult F 1 fi sh identifi ed to their stock of origin (Lake Averia and Lake Vanajanselk ä , correspondingly) by marking, showed genetic diversity comparable to that in their source populations (S1 and S7), indicating that genetic diversity was linear combinations of those of the contributing parental populations at the time when admixture occurs. In addition to some previous programs, it also takes into account genetic drift that has potentially occurred in parental populations and in admixture populations since the admixture, and even the potential drift of the parental populations before admixture occurred. The program additionally gives 95% confi dence intervals for the admixture proportions. Results of the STRUCTURE and LEADMIX analysis were compared with the population grouping analysis based on genetic distances. Analysis was carried out for four catch mixture populations: Lohjanj ä rvi 2000 (Table 1, S6), Vanajanselk ä 2000 (Table 1, S9) Ouluj ä rvi 1990s (Table 1, S12) and Ouluj ä rvi 2000 (Table 1, S13). For the Lake Lohjanj ä rvi case, the potential additional contributors were the populations from Lakes Painio, Averia and Vanajanselk ä , and for Lake Vanajanselk ä , the populations from Lakes Painio and Averia, (Fig. 3). For the Lake Ouluj ä rvi case, the only potential contributing foreign population was the Kivij ä rvi brood stock of Vanajanselk ä origin. RESULTS Genetic diversity in pikeperch samples The overall expected mean heterozygosity (H e ) of all samples was 0.51 and the mean allelic richness over all samples was 4.1 alleles loci ⫺1 . The F ST over all populations was as high as 0.08. The highest allele richness was observed in the Lake Kivij ä rvi broodstock population (S10), with 4.6 alleles loci ⫺1 , and highest expected Table 2. The genetic diversity of the pikeperch samples for the three cases, the sampling lake, catch year, status of the population, mean sample size for nine loci, mean expected heterozygosity ( H e ) and its standard error (SE), and average allelic richness (A r ) for nine DNA microsatellite loci. Sample Sampling lake Catch year Status Mean n/locus H e SE A r Case Lohjanj ä rvi S1 Broodstock, Averia, 1984 original/ source 58.0 0.41 0.08 3.7 S2 Lohjanj ä rvi 1994 – 2000 Averia, origin 56.6 0.46 0.08 3.9 S3 Broodstock, Painio 2006 original/ source 69.9 0.45 0.08 3.6 S4 Lohjanj ä rvi 1982 – 1986 original 59.1 0.44 0.09 3.6 S5 Lohjanj ä rvi 1994 – 2000 wild 59.4 0.45 0.08 4.0 S6 Lohjanj ä rvi 2002 – 2003 current, mixed 195.0 0.51 0.08 4.4 Case Vanajanselk ä S7 Vanajanselk ä 1988 – 1991 original 58.7 0.49 0.09 4.3 S8 Lohjanj ä rvi 1994 – 2000 Vanajanselk ä , origin 58.4 0.48 0.09 4.1 S9 Vanajanselk ä 2001 – 2006 current, mixed 204.2 0.53 0.08 4.5 Case Ouluj ä rvi S10 Broodstock, Kivij ä rvi 2008 original/ source 30.7 0.52 0.09 4.6 S11 Ouluj ä rvi 1946, 1974 original 14.6 0.54 0.07 3.8 S12 Ouluj ä rvi 1990s Mixed 95.3 0.46 0.10 4.0 S13 Ouluj ä rvi 2002 current, mixed 98.4 0.47 0.09 4.2 Mean 0.51 0.08 4.1 8 M. Salminen et al. Hereditas 149 (2012) Table 3. Mean proportions (%) of individuals assigned to each of the pre-defi ned populations from each of the four admixed pikeperch samples by the Bayesian STRUCTURE program (Bayes estimate), and stock proportion estimates with their 95% confi dence intervals (CI) from the maximum-likelihood estimation of the LEADMIX program (MLE estimate) . Bayes estimate % MLE estimate % 5% CI 95% CI S1 Lohjanj ä rvi 2000 Averia 6.4 10.2 0.0 20.9 S3 Painio 5.4 0.0 0.0 12.3 S7 Vanajanselk ä 1980 47.8 54.6 45.8 72.5 S4 Lohjanj ä rvi 1980 40.5 35.2 14.8 43.6 Vanajanselk ä 2000 S1 Averia 6.0 10.8 10.7 10.8 S3 Painio 2.2 3.1 3.1 3.2 S7 Vanajanselk ä 1980 91.8 86.2 86.1 86.2 Ouluj ä rvi 1990 S10 Kivij ä rvi-Vanajanselk ä 89.0 99.9 92.8 100.0 S11 Ouluj ä rvi, original 2.7 0.0 0.0 8.5 S12 Ouluj ä rvi 1990 8.3 – – – Ouluj ä rvi 2000 S10 Kivij ä rvi-Vanajanselk ä 89.8 98.3 89.9 100.0 S11 Ouluj ä rvi, original 6.5 1.7 0.0 11.8 S13 Ouluj ä rvi 2000 3.7 – – – in this case quite effi ciently transferred through rearing and stocking from the source lakes to the recipient population. Admixture analysis of three pikeperch populations Lake Lohjanj ä rvi In the Lake Lohjanj ä rvi case, a marked genetic contribution could be observed as result of the releases. The admixed catch sample was mainly composed of both Lake Vanajanselk ä and Lake Lohjanj ä rvi populations (Table 3, Fig. 4a). A large proportion of individuals were admixtures. Vanajanselk ä pikeperch comprised at least about half of the catch in Lake Lohjanj ä rvi. The maximum likelihood estimate for the contribution of Vanajanselk ä was 54.6% (95% confi dence interval (CI), 45.8 – 72.5%), and that of indigenous Lake Lohjanj ä rvi pikeperch was correspondingly only 35.2% (CI: 14.8 – 43.6%) (Table 3). The Bayesian estimate for the contribution of Vanajanselk ä was slightly less, being 47.8%, and for Lohjanj ä rvi somewhat more, 40.5%. In all cases, the Bayesian estimates were within the 95% confi dence intervals of the maximum likelihood estimates. The estimated contribution of the Vanajanselk ä population in the admixture was somewhat less than the observed contribution of marked Vanajanselk ä pikeperch (62 – 72%) to Lake Lohjanj ä rvi catch samples from year-classes 1993 – 1995 ( SALMINEN and RUUHIJ Ä RVI 2004). The most commonly released stock in Lake Lohjanj ä rvi has been Lake Averia pikeperch (Fig. 2). The latest releases have been carried out with Lake Painio pikeperch, but very little genetic effect of these releases could be observed, despite the relatively large numbers of released fi sh. The larger contribution from the Lake Painio pikeperch releases could be excluded (shown in Fig. 4a as green), but releases from Lake Averia were possibly contributing to the catch sample, as the maximum likelihood estimate was 10% with a probability interval of 0 – 20% for the Averia stock contribution. The more successful Lake Vanajanselk ä fi sh were only released in 1993 – 1995, and also in much smaller numbers. Lake Vanajanselk ä In the Lake Vanajanselk ä case, very little genetic contribution could be seen as a result of the releases (Fig. 4b), and the original Vanajanselk ä pikeperch still accounted for the majority of the catch. According to the Bayesian estimate, 91.8% of the admixture gene pool originated from the indigenous Vanajanselk ä population, and with the maximum likelihood estimation the proportion was somewhat less, being 86.2% (Table 3). About 10% originated from Hereditas 149 (2012) Genetic effects of supportive stockings on pikeperch 9 (Lohjanj ä rvi group), Lakes Vanajanselk ä , Ouluj ä rvi mixed samples, Kivij ä rvi and Lohjanj ä rvi samples the second group (Vanajanselk ä group) and the two northern native populations from Lakes Kemij ä rvi and Ouluj ä rvi the third group (Fig. 5). The genetic differentiation between allele frequencies of the populations before and after releases remained statistically signifi cant within all these groups, and in fact all pair wise comparisons between populations were statistically signifi cant. The mean genetic distance within the Lohjanj ä rvi group was 0.06 and within the Vanajavesi group 0.04. The original Lake Ouluj ä rvi sample was the most distinct of the other samples and grouped together with the sample of indigenous pikeperch from the northern Lake Kemij ä rvi (Fig. 1), which was used as an outgroup in the analysis. In general, the results on the genetic structure confi rmed the changes observed in the recipient populations in admixture analysis, and results from both analyses were congruent for all cases. Lake Lohjanj ä rvi Five of the six samples originating from Lake Lohjanj ä rvi pikeperch grouped into the same branch in the dendrogram (Fig. 5). The Averia 1984 sample was most similar to the original Lohjanj ä rvi sample, Painio had some unique features and the catch sample from the 1990s (wild fi sh identifi ed by marking) had also shifted somewhat from the original population. The most distinctive was, however, the last sample, the admixture from the years 2002 and 2003, which grouped into the Vanajanselk ä group instead of the Lohjanj ä rvi group. The F ST between the Lohjanj ä rvi the Averia population releases, and hardly any from the Painio population releases, although this population was the more commonly used stock in releases, and easily distinguishable from other populations. The 95% confi dence interval for the MLE estimate was narrow, and the Bayesian estimate did not fall within these limits. Lake Ouluj ä rvi In the case of Lake Ouluj ä rvi pikeperch, the indigenous population has nearly disappeared. The results show that a large majority of the contemporary Lake Ouluj ä rvi pikeperch population originated from the releases of Kivij ä rvi-Vanajanselk ä pikeperch. In both mixed samples, from 1990 and 2000, the proportion of fi sh originating from the releases was about 90% when estimated with the Bayesian method (Table 3, Fig. 4c – d). When the potential effect of genetic drift was included in the maximum-likelihood method, the proportion of the original Vanajanselk ä population increased very close to 100% (Table 3). For both Ouluj ä rvi mixed samples, the posterior probabilities were higher in the Bayesian method when the number of contributing populations was set to three rather than to two, indicating that the admixture could not completely be explained by the two populations, and the results are therefore given accordingly. Genetic differentiation among stocks before and after stocking According to genetic distances, the populations grouped into three main groups, where Lakes Averia, Painio and the original Lohjanj ä rvi samples formed the fi rst group Oulujärvi 1990s, mixed, S12 90 KEMIJ Ä RVI 2006, original OULUJÄRVI 1946, 1974, original, S11 91 60 Oulujärvi 2002, mixed, S13 42 VANAJANSELKÄ 1988–1991, original, S7 82 Vanajanselkä 2001-2006, current mixed, S9 36 Vanajanselkä 1994–2000, in Lake Lohjanjärvi, S8 Kivijärvi 2008, Vanajanselkä origin, S10 77 Lohjanjärvi 2002–2003, current mixed, S6 Averia 1994–2000, in Lake Lohjanjärvi, S2 Lohjanjärvi 1990s, wild, S5 54 58 58 78 Painio 2006, S3 Averia 1984, S1 LOHJANJÄRVI 1980s, original, S4 Fig. 5 . Genetic distances based on nine microsatellite DNA loci. Names of indigenous populations in capital letters. Sample numbers (S) refer to Table 1.