Microsatellites and multiplex PCRs for assessing ...

0 downloads 0 Views 537KB Size Report
Jan 25, 2014 - 1516, 95% CI: 422–10,000 in the James River in Virginia; Rose, 2008). ..... Lind, C.E., Evans, B.S., Knauer, J., Taylor, J.J.U., Jerry, D.R., 2009.
Aquaculture 426–427 (2014) 49–59

Contents lists available at ScienceDirect

Aquaculture journal homepage: www.elsevier.com/locate/aqua-online

Microsatellites and multiplex PCRs for assessing aquaculture practices of the grooved carpet shell Ruditapes decussatus in Spain Yaisel J. Borrell a, Alberto Arias-Pérez b, Ruth Freire b, Antonio Valdés a, José Antonio Sánchez a, Josefina Méndez b, Dorotea Martínez c, Jacobo López d, Carlos Carleos e, Gloria Blanco a, Ana M. Insua b,⁎ a

Departamento de Biología Funcional, Universidad de Oviedo, IUBA, 33006 Oviedo, Spain Departamento de Biología Celular y Molecular, Universidade da Coruña, 15071 A Coruña, Spain Centro de Cultivos Marinos de Ribadeo-CIMA, Xunta de Galicia, 27700 Ribadeo, Spain d Centro de Experimentación Pesquera, CEPEX, 33760 Castropol, Spain e Departamento de Estadística e Investigación Operativa y Didáctica de la Matemática, Universidad de Oviedo, 33007 Oviedo, Spain b c

a r t i c l e

i n f o

Article history: Received 28 October 2013 Received in revised form 10 December 2013 Accepted 16 January 2014 Available online 25 January 2014 Keywords: Ruditapes decussatus Microsatellite markers Multiplex PCR Diversity Parentage Effective breeding number

a b s t r a c t Supplementation aquaculture is intended to reinforce harvestable abundances of viable, naturally reproducing populations. The grooved carpet shell Ruditapes decussatus is one of the most important shellfish species in northern Spain (Asturias and Galicia), and their wild populations are annually supplemented using seeds produced in hatcheries. The current genetic status of these populations and a genetic evaluation of the consequences of the supplementation campaigns are lacking due to the absence of useful genetic markers that allow these kinds of studies. In this work, twelve variable microsatellite markers (mean HE = 0.663) and two useful multiplex PCRs are reported for R. decussatus. Different genetic characteristics were found between wild clams from Asturias and Galicia. Moreover, the seeds obtained in hatcheries for supplementation campaigns did not represent the wild gene pools well. Reductions of effective breeding numbers relative to the actual number of breeders were as large as 65%, due to unequal parental contributions and family variances. Finally, in an experimental supplementation programme conducted in a Galician population (Cambados), we report that the genetic status of the studied population changed significantly from one year to the next (FST = 0.011 P b 0.05) and we found what could be hatchery-produced seed (15%) in the wild restocked population. The accuracy of this estimate was evaluated using simulation procedures and we found less than 3% of type I error and values of 8–11% of type II error for three situations under analysis (32%, 10% and 1% of sampled true parent–offspring pairs) when using 95% as the threshold limit for parentage assignations. This work demonstrates the importance of temporal evaluations of the genetic status of supplemented and unsupplemented wild populations and indicates the need for changes in the protocols used for hatchery seed production for restocking purposes. A successful supplementation campaign can decrease genetic variance, and thus probably damage, the genetic status of wild populations. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Fishery enhancement aquaculture, or supplementation aquaculture, is intended to reinforce harvestable population abundances of viable, naturally reproducing populations (Utter and Epifanio, 2002). Usually, a fraction of the wild parents (or their offspring) is brought into captivity for reproduction or preferential survival, and the offspring are released into the natural habitat where they mix with wild conspecifics, although no exogenous genes are introduced (Ryman and Laikre, 1991). The target populations are often a resource that is highly valued by local communities, and the idea of assisting the wild population to support harvest pressures is rapidly accepted and funded. Concerns arise when potential impacts on recipient populations and aquatic communities outweigh any production benefits (Bert et al., 2007). The ideal ⁎ Corresponding author. E-mail address: [email protected] (A.M. Insua). 0044-8486/$ – see front matter © 2014 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.aquaculture.2014.01.010

expected result for a supplementation programme should fulfil the “enhancement function” but not compromise the future (not damage the evolutionary potential) of the target populations. Ryman and Laikre (1991) first brought this issue to attention and argued that in supplementation aquaculture some part of the overall population is favoured (in terms of survival). Favouring part of the population will result in an increase in family variances that consequently diminish the genetically effective population sizes (Nê) of the supported populations. Supplementation programmes can thus involve serious alterations in genetic diversity and decreases in fitness of the target populations (Bert et al., 2007). Utter and Epifanio (2002) reviewed supportive breeding practices for supplementation of wild marine species populations and found only one case (the red drum Sciaenops ocellatus) out of eight studied that seemed to fulfil (to some degree) the goals of a supplementation programme. More recently, Araki and Schmid (2010) reviewed the scientific literature on this subject from the past 50 years and suggested

50

Y.J. Borrell et al. / Aquaculture 426–427 (2014) 49–59

that scientific data supporting the positive effects of hatchery stocks on stock enhancement are largely missing. In shellfish, several studies trying to discern the best way/strategies for affording restoration programmes have been published recently (e.g., Lallias et al., 2010a). One of the key questions in supportive breeding practices is how large the contribution of the hatchery stock is to the breeding pool of wild individuals. It is supposed that the size of this contribution can be assessed using microsatellite markers (Araki and Schmid, 2010). Microsatellites are short, tandem-repeated sequences that are usually hypervariable and have proven to be valuable for research in different areas (Chistiakov et al., 2006). Microsatellites possess some drawbacks, such as null alleles or possible selection (e.g., Borrell et al., 2004; Nielsen et al., 2006); however, microsatellites combine the advantages of codominance, high polymorphism and multiple independently segregating loci. The use of microsatellites could describe population processes better than a single gene (Palsboll et al., 2007). Microsatellites are also useful tools for management strategies in aquacultured species because microsatellites help to determine parentage assignments and rates of inbreeding (Borrell et al., 2007, 2008; Herlin et al., 2008; Li et al., 2009; Lind et al., 2009; Navarro et al., 2008). Microsatellitebased parentage assignments have been proposed as a useful tool for the evaluation of reproductive fitness in natural settings, which is key for stock enhancement by hatchery-based stocking (Araki and Schmid, 2010; Araki et al., 2009; Boudry et al., 2002; Hedgecock et al., 2007; Lallias et al., 2010b). However, while parentage analyses within hatcheries have been undertaken with success (see above), parentage analysis in natural populations presents a valuable yet unique challenge because of the large numbers of pairwise comparisons, marker set limitations and the few true parent–offspring pairs sampled. These limitations can result in the incorrect assignment of false parent–offspring pairs that share alleles across multi-locus genotypes just by chance, biassing estimates of hatcheries' gene pools contribution to wild stocks. The use of strict exclusions, statistical thresholds and high numbers and quality of the markers have been proposed as the solution for undertaking with success this relevant task (Christie, 2010; Ford and Williamson, 2010; Harrison et al., 2013). The grooved carpet shell Ruditapes decussatus is distributed from southern and western England to the Iberian Peninsula and into the Mediterranean. R. decussatus is also present in southern to western Morocco and Senegal, West Africa (Poppe and Goto, 1991). R. decussatus is one of the most important shellfish species in Spain where fishing and consumption have been recorded since ancient times (e.g., 16th century). Between the years 1950 and 1990, global fishery production varied from 2000 to 4000 tons; however, in the early 1990s, 17,000 tons were captured in a single year (FAO, 2012a). The catch values plummeted in 1994 to only 3000 tons, and through 2010, the average annual catch did not reach 2000 tons worldwide (FAO, 2012a). Besides heavy fishing, clams have declined because of increases in pollution and growth in seaports and urban areas, thereby degrading their habitat. R. decussatus aquaculture, an activity that began in the 1980s, has produced an annual average of 4000 tons in the past 15 years, mainly from only a few countries. Portugal, Italy, France and Spain are currently the main producers. Global production appears to be declining; in 2010, aquaculture production reached only 2000 tons (FAO, 2012b). In Spain, clam fisheries are enhanced by supplementation of seeds from wild breeders that are induced to spawn in hatchery facilities, and then the seeds are released into the harvest areas. Curiously, a comprehensive evaluation of the success of this practice to truly enhance the exploited wild populations has not been conducted. Capture numbers by year are strictly controlled in all areas and are documented, but no other evaluations have been performed. Despite the importance of the analysis of genetic variability and population structure in managing exploited populations, the genetic status of R. decussatus populations in Spain is poorly studied. Available genetic markers to perform genetic studies in R. decussatus include some allozymes (e.g., Borsa et al., 1994), RAPDs (Pereira et al., 2011), introns

(Cordero et al., 2008; Gharbi et al., 2010), and mitochondrial loci (Gharbi et al., 2010), which are often characterised by a moderate level of polymorphism and/or low reproducibility. This work reports the identification of the first panel of microsatellite markers in R. decussatus, the development of multiplex PCR for quick and cheap genetic analyses within the species and a first evaluation of their utility for population analyses and parentage studies that could help improve supplementation strategies used in R. decussatus stock enhancement programmes. 2. Materials and methods 2.1. Samples A total of 348 individuals from wild populations in two northern Spain regions (Asturias and Galicia) were collected (Fig. 1). In Asturias, two sampling points were studied: Villaviciosa (46 individuals collected in April 2009; WVil09) and Eo (46 individuals, April 2009; WEo-09). In Galicia, samples from an experimental small area (approximately 4000 m2) located inside the Cambados's clams bed (2,000,000 m2), were collected in three consecutive years and were studied: samples from March 2009 (56 individuals; WCab-09), August 2010 (82 individuals taken just before the supplementation campaign conducted in 2010; WCab-10) and June 2011 (118 individuals taken in the supplemented area in 2010 with seeds from 2009; WCab-11). In addition, breeders extracted from those wild populations but long adapted to culture conditions in 2 hatcheries were induced to spawn. Samples from these hatchery broodstocks (n = 152) were analysed genetically. We did not count samples from all the breeders used for producing annual seeds within the hatcheries due to logistical limitations (Villaviciosa June 2009: 19 out of 100 breeders (19%) (BVil-09); Eo June 2009: 46 out of 150 breeders (31%) (BEo-09); and Cambados:

a)

b)

0.1

Fig. 1. a) Sampling locations in Spain and codes for the samples of R. decussatus analysed in this work using 12 microsatellite loci. Cab: Cambados-Galicia, Eo: Eo-Asturias, Vill: Villaviciosa-Asturias. Numbers refer to the year of sampling. b) Neighbour-Joining tree showing the unbiased Genetic Distance D (Nei, 1978) among R. decussatus samples analysed in this work. Samples from wild populations appear in circles, samples from breeders inside rectangles and samples from offspring are underlined. W: Wild, B: Breeders, O: Offspring.

Y.J. Borrell et al. / Aquaculture 426–427 (2014) 49–59

40 out of 57 breeders (70%) used in May 2009 (BCab-09) and 47 out of 49 breeders (96%) used in April 2010) (BCab-10). The sexes of the breeders were recorded for all the breeders in study, except the Eo breeders, using dissections and gonad inspections. The seeds produced in 2009 were sampled for genetic analyses (total n = 423) (Villaviciosa: 48 (OVil-09); Eo: 96 (OEo-09); and Cambados: 139 (OCab-09)). Samples from the seeds obtained in 2010 (Cambados: 140; OCab-10) were also sampled, although they were not used for supplementation campaigns. The seeds obtained in hatcheries in 2009 (OVil-09, OEo-09, OCab-09) were used for supplementation campaigns one year later (2010) in their respective areas of origin in Asturias and Galicia. In the Cambados's clam bed (Galicia), a small area, where density of wild clams is usually low, was used as a controlled zone for obtaining primary data about supplementation results. The supplementation was done in the 4000 m2 area, where we previously had sampled the wild population, using an approximate density of 20 seeds per square metre. 2.2. Development of microsatellite markers and multiplex PCRs Genomic DNA from one adult individual was extracted following the protocol described by Fernandez-Tajes and Mendez (2007). Microsatellite-containing sequences were obtained from an enriched genomic library constructed according to Billotte et al. (1999) using a biotin-labelled microsatellite oligoprobe (ATC)n and streptavidincoated magnetic beads. Recombinant clones were screened for repeats using the Tandem Repeats Finder software (Benson, 1999) and eightyone clones were sequenced using the ABI 3130XL Genetic Analyser. Sequences were aligned with BioEdit (Hall, 1999) using ClustalW for discarding similar sequences. A Blast procedure in the GenBank database was also performed. None of our sequences have been previously reported. Forward and reverse primers were designed for effective amplification of microsatellites using the FastPCR Professional package (Kalendar et al., 2009). Twenty-four microsatellite loci were amplified reliably in a sample of 40 wild individuals collected from Galicia and Asturias, northern Spain. This preliminary experiment allowed the collection of data regarding annealing temperatures and allelic size ranges. Fourteen microsatellite loci, available in the EMBL Nucleotide Sequence Database with accession numbers from HF565494 to HF565507 (Table 1), were arranged in two multiplex PCRs, RdMTP-1 (6 microsatellites) and RdMTP-2 (8 microsatellites), which were designed with the use of Multiplex Manager 1.0 software (Holleley and Geerts, 2009) (Table 1). Sample genomic DNA was purified from a small piece of abduct muscle tissues using the Zymobead™ Genomic DNA Kit (ZYMO Research, USA). The QIAGEN multiplex PCR kit protocol with an annealing temperature of 55 °C was used to prepare 15 μl of the two multiplex PCR amplifications. Individual genotypes were scored after analysing the amplification products on the ABI 3130XL Genetic Analyser using Genemapper 4.0. 2.3. Genetic diversity analyses The number of alleles at each microsatellite locus (NA), the proportion of heterozygous individuals (direct count heterozygosity, HO) and the unbiased estimate of heterozygosity (HE) for each group/sample were assessed using the Genetix software (Belkhir et al., 1996). The Fstat statistical package version 2.93 (Goudet, 1995, 2001) was used to estimate allelic richness (AR) and the total variation in gene frequencies (FIT) partitioned into components of variation occurring within (FIS) and among (FST) samples for each locus following the method described by Weir and Cockerham (1984). Significance levels for FIS were assessed by randomising alleles within samples 2000 times, followed by Bonferroni correction (Rice, 1989). Differences among groups of populations for a number of statistics (allelic richness, HO, HE, FIS, FST, relatedness (R) and corrected relatedness) were conducted using a two-sided statistical analysis included in the Fstat software. Relatedness is a

51

statistic described by Hamilton (1971) that is calculated in Fstat using an estimator strictly equivalent to the one used by Queller and Goodnight (1989). This statistical measurement will represent the average relatedness of individuals within samples when compared to the entire sample. All loci were also tested for linkage disequilibrium using Fstat. To check for genotyping errors, the data were analysed with the Micro-Checker software (Van Oosterhout et al., 2004). Tests for global population differentiation were applied using the loglikelihood G as the test statistic, with no assumption of random mating within samples (Goudet et al., 1996). Pairwise FST values between samples and p-values were also calculated using Fstat. To determine significance levels of FST, multilocus genotypes were randomised between pairs of samples (2000 permutations); then, the significance after Bonferroni correction was calculated (Rice, 1989). Arlequin 3.5 was used to partition genetic variability among samples using locus-bylocus AMOVA (Excoffier et al., 1992, 2005). This procedure is recommended when using unlinked loci (L. Excoffier, pers. comm., the Genetic Software Forum). The unbiased genetic distance (D) among samples (Nei, 1978) and the neighbour-joining clustering method included in the PHYLIP 3.5p package (Felsenstein, 1993) were used to generate trees that were visualised using the TreeView programme (Page, 1996). 2.4. Estimating effective breeding numbers (Nê) The rate of inbreeding (ΔF) and the effective population size (Nê) are related as ΔF = 1 / (2Nê) (Falconer, 1989). We estimated effective population sizes within hatcheries using the classical formula for estimating Nê: Nê = 4 x (Nm × Nf) / (Nm + Nf) (Falconer, 1989). We also used Chevassus's formulation for estimations of Nê considering unequal contributions (family variances): Nê = 4 (N − 2) / ((Kd + Vd/Kd) + (Ks + Vs/Ks) − 2), where N is number of offspring; K is the average number of offspring per parent; V represents variances of the number of offspring per parent; d represents dams and s represents sires (Chevassus, 1989). 2.5. Parentage assignments We used Cervus 3.0 software (Kalinowski et al., 2007; Marshall et al., 1998) to assign parentage. This programme calculates both the a priori polymorphic information content (PIC) for every locus from each broodstock and the total exclusionary power (E). In addition, the programme simulates parental assignments. Cervus uses simulation of parentage analysis to evaluate the confidence in assignment of parentage to the most likely candidate parent. Parentage analysis is carried out with the simulated genotypes as it is with real genotypes, but in the simulation the identity of the true parent is known for each offspring. Cervus compares the distribution of LOD or Delta scores for tests in which the most likely candidate parent is the true parent with the distribution of LOD or Delta scores for tests in which the most likely candidate parent is not the true parent. Confidence in assignment is defined as the proportion of all candidate parents with LOD or Delta scores exceeding a given LOD or Delta score that are true parents (Cervus 3.0, Marshall et al., 1998). The parentage assignment simulations within hatcheries were conducted by considering the total number of breeders and the percentages of sampled breeders per broodstock. Ten thousand cycles of simulated assignments were performed using Cervus confidence intervals. Finally, after genotyping, all the offspring were assigned to the most likely candidate parent pair. In the assignment procedures, we allowed for typing errors (0.01), as this dramatically reduces the impact of two other possible causes of mismatches in parent–offspring relationships: mutations and null alleles (Kalinowski et al., 2007; Marshall et al., 1998). Parentage assignment was also used for assessing supplementation's consequences in the wild. In the wild sample obtained in Cambados in 2011 (WCab-11), offspring from the breeders from 2009 (BCab-09) could appear due to the supplementation process conducted in 2010.

52

Table 1 Microsatellites characteristics, multiplex PCRs and results after genetic analyses in R. decussatus individuals coming from thirteen Spanish samples. N

NA

AR (N = 19)

Size range (bp)

HO

HE

PIC

RdATC-223

922

11

4.68

85–115

0.370

0.405

0.383

RdATC-219

915

8

3.25

216–237

0.436

0.476

RdATC-199

919

9

5.47

164–188

0.743

RdATC-185

920

11

4.22

128–159

RdATC-238

828

13

7.16

RdATC-263

916

12

RdATC-28b

922

RdATC-212b

FIT

FST

FIS

B

0.090

0.037

0.055

0.042

0.397

0.086

0.016

0.071

−0.042

0.749

0.707

0.011

0.028

−0.018

0.388

0.404

0.376

0.042

0.035

0.007

122–164

0.588

0.825

0.801

0.290

0.036

0.263*

4.43

177–207

0.592

0.643

0.583

0.084

0.050

0.036

8

3.45

184–205

0.376

0.442

0.386

0.151

0.016

0.137*

0.098

(TGA)3TCA(TGA)6T(TGA)2TAA(TGA)4 ATA(AGA)3(TGA)3AGAGGG(GGA)4 CGA)3 (TGC)4..... (TGG)3

920

10

6.20

230–257

0.735

0.759

0.721

0.036

0.032

0.004

0

(ATG)4… (ATG)11

RdATC-022

920

9

5.84

131–155

0.739

0.754

0.714

0.023

0.026

−0.004

0.068

(ATG)4ACG(ATG)8

RdATC-215

914

19

9.78

104–185

0.882

0.856

0.841

−0.029

0.010

−0.040

−0.006

RdATC-1.34

890

11

7.71

229–256

0.731

0.807

0.785

0.096

0.023

0.075*

−0.064

RdATC-177

869

16

8.89

225–267

0.793

0.850

0.832

0.068

0.012

0.057*

−0.009

RdATC-125

922

12

7.11

143–176

0.735

0.725

0.694

−0.013

0.018

−0.032

0.064

RdATC-1.79

890

15

9.26

301–343

0.842

0.837

0.819

−0.003

0.020

−0.023

−0.015

Averages:

905

11.71

6,25

0.639

0.681

0.646

0.064

0.025

0.040*

0.046 −0.036 0.116* −0.002

Motif

Primer Pairs

Multiplex PCR

Dye

AN

(ATG)7

F: AGCATGCTGAGAGAATGTTG R: CTCCAAGAGCTTTGCAGTCA F: GATGATTGACATGGTTGATGAC R: CCCCTAACCCTGCTAGATTTG F: AAAAGTCCGGAATACGCAGA R: CGGTACCTTTCCTCTCTTGG F: TATGGTCATTGCGGACTTGA R: CGCGTTAGCATCATCTGAAA F: CCATGTAGACAGTGATCCTTCG R: CAGCCTTCTCCTCCTCATCA F: GATGCTGTTGTCGCAGTTGT R: AGCTAGGTTCGTGCCTTATGA

RdMTP- 1

6-FAM

HF565494

RdMTP- 1

6-FAM

HF565495

RdMTP- 1

VIC

HF565496

RdMTP- 1

VIC

HF565497

RdMTP- 1

PET

HF565498

RdMTP- 1

NED

HF565499

F: GTGGTGCAAACGTTGATGTG R: ATCGTCATCACCGTtGTCGT F: ATCGCGTTTCTGCTCGTAAT R: CGACCGTAAAGTCACACCTG F:AAAGAAAGTCCGGTATGTCCA R: ACCTTTCCTCTCTTGGTCAGT F: ATGCAACGGCTAAATCTTGG R: CCGGCTAGGGAAACAATGTA F: CTTCAGCAGAATTATCAAGTTCCG R: AGTATCAGTTCTTGTCAAGGATGACG F: GCTCAGTTTGGTTGCTCATA R: CACATTTGCAATAGCTGTCT F: ACTTTTTACGGCAGCCACAC R: AATCGGGATTTTGATGATGG F: TTAAAGTTGTTGCACTAAGCAGAAC R: CGCAAATTCTTCGCCTTTAT

RdMTP- 2

6-FAM

HF565500

RdMTP- 2

6-FAM

HF565501

RdMTP- 2

VIC

HF565502

RdMTP- 2

PET

HF565503

RdMTP- 2

PET

HF565504

RdMTP- 2

NED

HF565505

RdMTP- 2

NED

HF565506

RdMTP- 2

NED

HF565507

(TGA)4…(TGA)2GAATGACAG(TGA)5… (TGA)3TCA(TGA)5 (AAAG)3… (ATG)13 (TGA)2(TGT)3(TGA)7 (ATG)14

(TGT)5TAT(TGT)2(TGA)3AGATGG (TGA)3TGGCGG(TGG)2(TGA)9 (ATG)5G(ATG)2G(ATG)10(ATT)7 (ATT)7ATG(AAG)2AGG(ATT)2AATAGT (ATT)2(ATC)16 (ATC)9 (GAT)14

N: sample sizes. NA: number of alleles per locus. AR: Allelic richness for the minor possible number of diploid individuals by sample (19). HO: Observed Heterozygosity. HE: Expected Heterozygosity. PIC: Polymorphic Information Content. Weir and Cockerham (1984) F statistics: FIT, FST, FIS (significance evaluated using 2000 permutations in Fstat software). B: Brookfield 1 statistic for null allele's inferences (*: q N 0.05) using the Microchecker software. * P b 0.05. AN: Accession numbers at the EMBL Nucleotide Sequence Database. Two loci are underlined since they were not taken into account for further analyses because of evidences of possible null alleles (RdATC-238) and of evidences of possible linkage disequilibria (RdATC-022) (see the text for further details).

Y.J. Borrell et al. / Aquaculture 426–427 (2014) 49–59

Locus

Y.J. Borrell et al. / Aquaculture 426–427 (2014) 49–59

Hatchery parent–pairs for sample WCab-11 were then searched among the hatchery parents (BCab-09) using Cervus. However, parentage analysis in the wild is a different problem and it needs a different approach since it can result in the incorrect assignment of false parent–offspring pairs. An evaluation of the associated errors in the parentage analysis is needed. It is also needed to evaluate the effective population size of the population under study since this can strongly influences assignment results. Using Cervus's simulations is not possible to control or to know clearly the parent crossings (it can include crossings between wild and hatchery parents) or/and the constitution of the resultant offspring population. We then carried out another simulation approach. We simulated progenies (and their genotypes) from WCab-09 (56 parents) and BCab-09 (40) parents using the programme ProbmaxG (Danzmann, 1997). A total of five, twenty, and finally two hundred virtual offspring per full-sib family from full factorial crosses of 28♀ × 28♂ in WCab-09 (sexes were unknown) were obtained (3920, 15,680 and finally 156,800 wild offspring). These populations with wild origins were then mixed with offspring coming from crossings among hatchery parents. Five offspring per full-sib family from crossing of 24♀ and 16♂ (BCab-09) were obtained (1920 offspring with hatchery origin). Mixed populations showed then different characteristics in terms of percentages of hatchery individuals/true parents–offspring pairs sampled (32.9%, 10.9% and 1.2%). For avoiding potential sampling effects of particular parent combinations which may appear in our previous approach we carried out another simulation approach for generating progenies. We simulated progenies (and their genotypes) from WCab09 (56 parents) considering that the wild population is in Hardy– Weinberg equilibrium and then we can assume panmixia. Therefore, wild offspring were generated by random sampling of alleles in accordance with allelic frequencies in the wild parental population. Also, linkage equilibrium was assumed so alleles from different microsatellites were independently sampled. The code used to compute allele frequencies and generate offspring genotypes is available at ftp://carleos.epv. uniovi.es/panmixia/. Parent–pairs for all of these virtual offspring were then searched among the hatchery parents (BCab-09) using Cervus. We could evaluate then type I (wild individuals incorrectly classified as individuals with hatchery origins) and type II errors (offspring with hatchery parents included in the study but not assigned) in the parentage analyses. Finally, the software Colony (Wang and Santure, 2009) was used. It estimates the current effective population size (Nê) from sibship assignments (Wang, 2009). A small population (small Nê) will contain a high proportion of sibs because the smaller the Nê, the greater the probability that two individuals drawn at random from the same cohort within a population are sibs that share one parent or both parents. Three replicates of short runs (Wang and Santure, 2009) using polygamy (both parents) and the full likelihood method (Wang, 2004), with no genotyping errors at the markers allowed, were performed.

53

3. Results 3.1. RdMTP-1 and RdMTP-2 multiplex PCRs and the genetic characterisation of wild, breeder and seed individuals Fourteen microsatellite loci were developed for the R. decussatus species (Table 1) and arranged into two multiplex PCRs that work well and have a mean proportion of individuals typed per locus of 0.9781. The electrophoretograms were clear and lacked noise or artefact peaks, thereby allowing easy and fast genotyping of a high number of individuals. The fourteen microsatellite loci developed showed medium/high levels of genetic variation in the assayed samples with ranges of 8 to 19 alleles per locus, and the expected and observed heterozygosities ranged from HO = 0.370 (RdATC-223 locus)-HE = 0.404 (RdATC185 locus) to HO = 0.882-HE = 0.856 (RdATC-215 locus). The genetic loci included in RdMTP-1 and RdMTP-2 multiplexes have PIC mean values of PICRdMTP-1 = 0.541 and PICRdMTP-2 = 0.723, and the global mean is PICRdMTP-1 + RdMTP-2 = 0.646. The multiplexes could correctly identify true parent pairs with known sexes (following 10,000 simulation steps in Cervus 3.0 using 50 genotyped breeders) in RdMTP-1, RdMTP-2, and RdMTP-1 + RdMTP-2 in 53.0%, 97.8%, and 99.8% of the progenies studied, respectively. After obtaining genetic data for the samples, we found evidence for high probabilities of occurrence of null alleles for the RdATC-238 locus (Brookfield 1 Statistic = + 0.116) (Table 1). In addition, we found significant linkage disequilibrium evidence (P b 0.000042 in all the samples) between the RdATC-199 and RdATC-022 loci. Following this result, we eliminated the RdATC-238 (null alleles) and RdATC-022 loci from further genetic diversity analyses and parentage inferences. The RdATC-022 locus was eliminated to maintain balance in the multiplex PCRs in terms of the number of loci included in each; the resulting multiplex PCRs had 5 (RdMTP-1) and 7 (RdMTP-2) useful loci (12 microsatellite loci in total) (Table 1). The final, selected twelve microsatellite loci could correctly identify true parent pairs for 10,000 progenies studied after simulations with 50 and 100 parents with known sexes in 99.2% and 98.7% of the cases, respectively. The genetic characteristics of all the samples analysed here are presented in Table 2. No significant differences in genetic variation levels between regions (Galicia–Asturias) for samples of wild populations were found. Significant differences in variability were not found between wild and breeder samples within localities. However, seeds showed more significant minor allelic richness values (AR = 5.27) than breeders (AR = 6.04) and wild populations (AR = 6.16) in a global analysis (P = 0.0014). Additionally, the global seeds FST (FST = 0.045) and relatedness values (R = 0.087) were significantly higher (P = 0.03 in both cases) than their breeder (FST = 0.012, R = 0.023) and wild (FST = 0.005, R = 0.010) counterpart values. Four (Villaviciosa

Table 2 Genetic variation data after the genetic analyses in R. decussatus (925 individuals) from Spain. Sampling site

Villaviciosa, Asturias

Eo, Asturias

Cambados, Galicia

Sample

Wild 04/2009 Breeders 06/2009 Offspring Seeds 05/2010 Wild 04/2009 Breeders 06/2009 Offspring seeds 05/2010 Wild 03/2009 Wild 08/2010 Wild 06/2011 Breeders 05/2009 Offspring Seeds 08/2010 Breeders 04/2010 Offspring Seeds 05/2011

Code

WVil-09 BVil-09 OVil-09 WEo-09 BEo-09 OEo-09 WCab-09 WCab-10 WCab-11 BCab-09 OCab-09 BCab-10 OCab-10

12 loci N

NA

AR

HO

HE

FIS

46 19 48 46 46 96 56 82 118 40 139 47 140 923

7.75 6.17 6.33 7.83 7.42 6.42 7.50 7.92 9.25 7.33 7.33 7.17 6.67 11.83

6.19 6.07 5.49 6.10 6.08 4.93 5.86 5.80 6.83 6.11 5.32 5.88 5.33 6.23

0.675 0.630 0.651 0.621 0.674 0.668 0.608 0.635 0.634 0.631 0.608 0.646 0.621 0.638

0.664 0.658 0.670 0.652 0.665 0.619 0.627 0.652 0.734 0.651 0.613 0.656 0.620 0.663

−0.018 n.s 0.044 n.s 0.029 n.s 0.048 n.s −0.014 n.s −0.080 n.s 0.030 n.s 0.026 n.s 0.136 * 0.031 n.s 0.008 n.s 0.015 n.s −0.003 n.s 0.031 *

N: sample sizes. NA: allele numbers per locus. AR: Allelic richness for the minor possible number of diploid individuals by sample (19). HO: observed Heterozygosy. HE: Expected heterozygosity. FIS: degree of departure from expected Hardy–Weinberg proportions within samples. *P b 0.00032. n.s = Not significant.

54

Y.J. Borrell et al. / Aquaculture 426–427 (2014) 49–59

Table 3 Pairwise genetic differentiation between Spanish samples of R. decussatus after genetic analyses using 12 microsatellite loci. FST statistics, P values (between parenthesis) and its significance after Bonferroni corrections at an alpha level of 5% (P = 0.0006). Sample

WVil-09

BVil-09

OVil-09

WEo-09

BEo-09

OEo-09

WCab-09

BCab-09

OCab-09

WCab-10

BCab-10

OCab-10

WCab-11

WVil-09

-

0.004 n.s (0.1057)

0.023 * (0.0006)

0.000 n.s (0.9012)

0.003 n.s (0.0775)

0.047 * (0.0006)

0.016 * (0.0006)

0.011 n.s (0.0275)

0.025 * (0.0006)

0.012 * (0.0006)

0.008 n.s (0.0064)

0.027 * (0.0006)

0.017 * (0.0006)

-

0.005 (0.5756)

0.001 n.s (0.1153)

0.003 n.s (0.1647)

0.056 * (0.0006)

0.016 n.s (0.0371)

0.010 n.s (0.0794)

0.024 * (0.0006)

0.009 n.s (0.1717)

0.015 n.s (0.0019)

0.031 * (0.0006)

0.015 n.s (0.0038)

-

0.014 n.s (0.0006)

0.012 * (0.0006)

0.061 * (0.0006)

0.025 * (0.0006)

0.019 * (0.0006)

0.030 * (0.0006)

0.021 * (0.0006)

0.027 * (0.0006)

0.041 * (0.0006)

0.020 * (0.0006)

-

0.000 n.s (0.1025)

0.045 * (0.0006)

0.014 n.s (0.0051)

0.007 n.s (0.0128)

0.023 * (0.0006)

0.009 n.s (0.0051)

0.007 n.s (0.0012)

0.021 * (0.0006)

0.018 * (0.0006)

-

0.032 * (0.0006)

0.009 n.s (0.0230)

0.005 n.s (0.0025)

0.016 * (0.0006)

0.007 n.s (0.0012)

0.006 n.s (0.0025)

0.019 * (0.0006)

0.013 * (0.0006)

-

0.060 * (0.0006)

0.047 * (0.0006)

0.054 * (0.0006)

0.047 * (0.0006)

0.052 * (0.0006)

0.072 * (0.0006)

0.054 * (0.0006)

-

-0.003 n.s (0.2141)

0.003 n.s (0.0025)

0.001 n.s (0.7987)

0.006 n.s (0.0076)

0.016 * (0.0006)

0.015 * (0.0006)

-

0.003 n.s (0.3480)

-0.004 n.s (0.9666)

0.000 n.s (0.8512)

0.013 n.s (0.0012)

0.013 n.s (0.0012)

-

0.007 * (0.0006)

0.013 * (0.0006)

0.021 * (0.0006)

0.023 * (0.0006)

-

0.000 n.s (0.0653)

0.015 * (0.0006)

0.011 * (0.0006)

-

0.004 n.s (0.7262)

0.011 * (0.0006)

-

0.029 * (0.0006)

BVil-09 OVil-09 WEo-09 BEo-09 OEo-09 WCab09 BCab-09 OCab-09 WCab10 BCab-10 OCab-10

W: Wild, B: Breeders, O: Offsprings. Vil: Villaviciosa, Asturias. Eo: Eo, Asturias. Cab: Cambados, Galicia. *: P