Final Report Summary - REPROSEL (REPROduction protocols and molecular tools for mass spawning and communal rearing based SELective breeding schemes applied to multiple-spawning marine fish)
Selective breeding schemes are widely recognised as efficient means to improve aquaculture production. However, fish farmers are reluctant to invest in such programs due to the required investments (space, labour and maintenance) and technical know-how. As a shortcoming solution mass selection schemes are often implemented notwithstanding noticeable drawbacks able to void the genetic gain. In REPROSEL, these drawbacks are tackled with the aim of making mass spawning a viable way to implement selective breeding schemes in the European seabass (Dicentrarchus labrax) and the Gilthead seabream (Sparus aurata).
We describe the daily and weekly spawning kinetics of cultured broodstock analysing the family structure of single spawning events (45 for the two species) over the entire reproduction period. A further analysis of the parental contribution of hormonally induced commercial broodstock (16 batches of 200 offspring) allowed developing protocols and guidelines on mass spawning synchronisation. New relative to the size of the broodstock appeared to be affected by the hormonal induction but also by the sex ratio, the number of breeders and the size of the tank.
Public databases were screened to develop ultra-efficient microsatellite (STR) and single nucleotide polymorphism (SNP) panels for parentage assignment. 13 seabream and 14 seabass STRs mapped in different linkage groups were selected to be amplifiable in a single tube. Each multiplex, first tested on known families, was validated on the farms' breeders (288 seabream and 235 seabass). SNPs were mined from the Sigenae database. They were detected based on the number of software supporting them with constraints on depth, minimum allele frequency, flanking sequences and GC%, and further annotated and positioned on the stickleback linkage map with ENSEMBL. From more than 400 identified SNPs per species, 128 were submitted for array synthesis (AB). 69 seabream and 61 seabass SNPs were validated as highly efficient for parentage assignment. We show that the assignment power of a large STR multiplex can be outperformed by an optimised SNPs panel having five times more markers.
A new allocation software was developed. It uses the exclusion method as the first approach to allocate the parentage. In case of multiple allocations, it uses the likelihood approach. The program follows a stepwise procedure allowing the progressive exclusion of markers, which appears to be useful for SNPs due to their higher global risk of mismatch. Common ±2 pb errors and larger genotypic discrepancies remain distinguishable. Two input files, parents and offspring, are required and can be easily generated by a simple copy / paste operation from any tab delimited text file. The main outputs are the parental allocations with their probability estimation, the crossing scheme, the observed and expected heterozygosities and the PIC and exclusion probabilities.
Breeding designs were optimised by comparing obtained rate of inbreeding and genetic gain using a stochastic simulation model of the breeding program. The simulations took into consideration the number, sex-ratio, random mating and contribution of breeders and the number of broodstock tanks. A stepwise approach is proposed to genotype and select groups of fish based on their pedigree and BLUP breeding values, so contributing to limit the genotyping costs. A tool that puts a cost-factor on inbreeding to the estimated breeding values has been developed to control the rate of inbreeding. By applying the suggested breeding schemes, it is possible to achieve a genetic gain in growth of at least 10 % per generation.
Project context and objectives:
Selective breeding programs require long-term investment, and their results depend on the compromise between the genetic response obtained and the maintenance of the genetic variability, i.e. controlling rates of inbreeding. These two fundamentals may be put at risk when choosing separate rearing of full and half-sib groups produced by different parents, in situations where:
(a) significant between-tank effects increase the phenotypic variability affecting the genetic value estimates of sib groups; and
(b) lack of reproduction control results in a reduction in the number of families produced at each generation and skewed contributions of parents, resulting in increased rates of inbreeding.
The above risks are especially high for species characterised by batch-spawning strategies. Moreover, the man-power, facilities and broodstock required for an effective breeding program based on separate family rearing is a strong discouraging factor for SMEs to commit to family-based breeding programmes.
In salmonids, breeding schemes are based upon the manual stripping of mature gametes from the chosen broodfish and the artificial fertilisation of these gametes. For other species like the European seabass and the Gilthead seabream, characterised by batch-spawning strategies, the outcome of artificial fertilisation is awkward. The main limitations in the females derive from:
(a) the difficulty to synchronize the ovulation and spawning of numerous females in a single spawning event;
(b) the failure to predict accurately the time of ovulation in order to obtain the eggs at a prime condition;
(c) the very brief period after ovulation (1 -3 hours) before over-ripening can reduce egg viability to 0 %; and
(d) the limited quantity of good quality eggs available from one mature female in the case of the Gilthead seabream.
In these conditions, the optimisation of mass spawning to produce adequate pools of families and the development of competitive molecular panels for parentage assignment may become a valid alternative to implement selective breeding programs in these species. The main objective of REPROSEL is to provide the participating SMEs with tools to implement breeding programs with constrained rates of inbreeding, but without the necessity of utilising artificial insemination.
Batches of progenies will be obtained first from mass spawning of commercial broodstock raised in traditional production facilities to depict the parental contribution to the new generation, an opportunity which will open new strategies for broodstock management. Other progeny batches will be produced in experimental facilities using smaller broodstock tanks in order to better understand the spawning kinetics over the whole production season. Relying on mass spawning will eliminate the stress and potential damage of broodfish ensuing from the stripping operations and will avoid the disruption of production routine and the financial costs that it entails.
STR panels are already available for parentage inference on both species but most often their power does not bring to the high level of parentage assignment expected in commercial selective programmes. Consequently, new highly discriminative molecular panels for parentage assignment will be developed using STR and SNP markers, the latter being seen as a valid alternative to STRs. Their efficiency and costs will be compared.
The efficient sibship reconstruction, allowing a high and reliable level of unambiguous parentage assignment, is essential to build-up on the higher power of the markers panels. A new allocation software, combining the exclusion and probabilistic approaches, will be developed and adapted to the specificity of SNP markers. The objective is also to deliver an easy-to-use tool to prepare the input files and interpret the output information.
Families derived from a mass spawning event will keep maintaining an unfavourable structure for the needs of selective breeding, even if improvements are expected from the better understanding of the reproduction patterns of the species. Stochastic simulations will be used to account for such limits and propose breeding schemes adapted to the participating SMEs. Lack of identification of individuals and skewed contributions of parents will be compensated by an increased total number of individuals in order to assure wanted family sizes and increased number of batches. The different selection schemes to be explored will have controlled rates of inbreeding. These are two factors which reduce the risk of the breeding program, and together with multi-trait selection assure sustainable breeding programs for the SMEs.
With the proposed approach, the separate rearing of sibling groups will be replaced by the rearing in one or few communal tanks. The possibility to start the breeding evaluation using one or few tanks constitutes a considerable progress in term of required facilities and associated costs, and this will facilitate the access of SMEs to the proven benefits of selective breeding programmes. This approach is also expected to reduce undesired initial tank effects and the risks edging on the genetic evaluation inherent to the confusion between genetic and non-genetic effects.
Project results:
WP2
The aims of WP2 were to:
- produce post-larvae and fry batches from captive commercial and experimental seabass and seabream broodstock,
- describe the spawning kinetics and parental contribution of captive broodstocks and increase parental contribution in commercial hatcheries using hormonal manipulations to synchronize spawning,
- develop and propose a protocol for the increase of the number of spawning parents and equalise parental contribution, to be adopted by the industry.
Task 2.1. Description of spawning kinetics and parental contribution
Associated deliverables:
D2.2: Biological sampling of broodstock
D2.3: Synchronisation of spawning and production of fry batches (year 1)
D2.4: Synchronisation of spawning and production of fry batches (year 2)
D2.5: Description of spawning kinetics and endocrine control of seabass and seabream broodstock
D2.6: Development of spawning protocols for optimised parental contribution
D2.7: Guidelines of spawning protocols for optimised parental contribution.
IRTA set up and produced the required larval and blood samples from three tanks of European seabass broodstock and two tanks of gilthead seabream. Each of the four SMEs set up broodstock groups that were induced with the identified hormone protocols to produce the required seabass (TNSL and CUPIMAR) and seabream (ARDAG and GMF) larval samples. Broodstock fin-clips, larval and fry samples were collected.
The European seabass spawning kinetics were described (D2.5) in two tanks of cultured broodstock held in IRTA. Each tank had a biomass of 28 kg of seabass, tank A had five males and five females and tank B had four females and six males. Tanks A and B produced respectively, 11 and 15 spawns with approximate fecundities of floating eggs over the entire spawning season of 75 000 eggs.kg-1 and 100 500 eggs.kg-1. Both tanks started the spawning season with a single natural spontaneous spawn from a single female fertilised by three males. After this initial spawn the females were induced to spawn every 10 days with a single injection of 10 µg.kg-1 GnRHa (gonadotropin releasing hormone agonist), whilst males were implanted once with a slow release implant, 20 µg.kg-1. The hormone inductions produced four spawning periods in each tank, one induction in tank B produced three spawns, five inductions produced two spawns and two inductions, the last successful spawning in each tank was a single spawn. Both tanks had dominant males that participated in most spawns analysed, in tank A one male fertilised 73 % (participating in all spawns) of the larvae analysed and in tank B, two males fertilised 43 % and 47 % of the larvae analysed. All other males fertilised 0 % to 12 % of the larvae. Female contributions were also unequal, in tank A, two females contributed 45 % and 42 % of the larvae, one female 13 % and two females did not contribute. In tank B, one female contributed 66 %, two females 18 % and 16 % and one female did not contribute. One female in tank B contributed to all spawning periods, the initial spontaneous spawn and all four inductions to give a total of five spawns. Generally all spawns were from a single female (69 % of spawns) fertilised by three males (46 %) although the participation of just one (15 %), two (15 %) and four (15 %) males was also observed. The mean number of families obtained from a spawn was 3.2 ± 1.4 families and the mean number of families obtained from a hormone induction (usually two spawns) was 5.3 ± 1.2 families.
Plasma profiles of hormone changes (D2.5) exhibited clear peaks in estradiol (3.71 ± 0.83 ng.mL-1) testosterone (6.26 ± 2.61 ng.mL-1) and 11-ketotestosterone (2.67 ± 0.38 ng.mL-1) during 2012, when seabass were observed to spawn. No peaks and approximately half the plasma hormone level were observed during 2011, which presented poor spawning. Levels in wild and cultured broodstock were similar indicating that any reduction in spawning in cultured stocks was not due to significant decreases in hormone profiles.
The gilthead seabream spawning kinetics were described (D2.5) in two tanks of cultured broodstock held in IRTA. Each tank had a biomass of 34 - 35 kg of seabream, which consisted of 5 males and 7 females. Both tanks produced daily natural spontaneous spawning, tanks C1 and C2 produced respectively 121 and 120 spawns with approximate fecundities of floating eggs over the entire spawning season of 1.2 million eggs.kg-1 and 0.9 million eggs.kg-1. The spawning in both tanks lasted over four months from late December / early January to late May. Spawns were collected and analysed at approximately weekly intervals during the first two months of spawning and the five to six days of consecutive daily spawning were analysed from the fifth week. Both tanks had dominant males that participated in all spawns analysed, in C1 one male fertilised 91 % of the larvae analysed and in tank C1, one male fertilised 67 % and a second male 25 % of the larvae analysed. Both tanks also contained a male that did not participate in any spawning. Female contributions were more even and in C1, the contribution of four females varied from 15 % to 37 %, while three females contributed from 0 % to 2 %. In tank C2, the contribution of three females varied from 24 % to 46 %, while four females contributed from 0 % to 2 %. During the period of consecutive days of spawning, both females and males were observed to spawn daily. In tank C1, two females and one male spawned every day during five days and tank C2, one male spawned every day during six days and another male and two females spawned on five of the six days. Generally most spawns were contributed to by two males and two females and the mean number of families obtained from a spawn was 5.6 ± 2.5 families. There was a tendency for the number of families to increase from the early spawns in the season to the last spawns analysed that coincided with the end of the second month and the start of the third month of the 4+ month spawning season.
Task 2.2. and 2.3: Synchronisation of spawning in European seabass and Gilthead seabream
Associated deliverables: D2.2 D2.3 D2.4 D2.6 D2.7
Spawning protocols were developed to optimise the parental contribution of spawning European seabass (D2.6). In Cupimar, two tanks of wild seabass were used, one tank was a control tank from which natural spontaneous spawns were collected and the second tank was hormonally induced, the females were induced to spawn with a single injection of 10 µg.kg-1 GnRHa, whilst males were implanted once with a slow release implant, 20 µg.kg-1 GnRH. The control tank had 13 males and six females with a total biomass of 51 kg. The hormone tank had nine males and nine females and two unidentified fish with a total biomass of 56 kg. The volume of eggs obtained from the two tanks was similar, the control gave 1.6 L of floating eggs, which was approximately 23 250 eggs.kg-1 and the hormone group gave 1.4 L of floating eggs, which was approximately 19 500 eggs.kg-1. In the control group, four females and seven males participated to give 15 - 17 families. The participation of individuals increased slightly in the hormone group and six females and seven males participated, but the number of families (13 - 15) was slightly lower. In TNSL, the same large group of 116 cultured seabass (size tank) was hormonally induced to spawn over consecutive years. The group consisted of 45 females, 62 males and 9 unidentified fish that had a biomass in 2011 of 232 kg and in 2012 of 330 kg. In 2011, just the females were induced using a single injection of 10 µg.kg-1 GnRHa, whilst in 2012 the females were given the same hormonal induction and the males were implanted once with a slow release implant, 20 µg.kg-1 GnRHa. In 2011, 13.7 million floating eggs were collected, which represented approximately 58 900 eggs.kg-1 and in 2012 8 million floating eggs were collected, which was approximately 24 300 eggs.kg-1. Two spawns were analysed from each year (four spawns in total) and the parental contribution was similar with a mean participation of 24 ± 8 males and of 17 ± 2 females to produce a mean of 41 ± 10 families. The large group of broodstock combined with hormone induction appeared to increase the number of families compared to the smaller groups in Cupimar and IRTA.
Spawning protocols were developed to optimise the parental contribution of spawning gilthead seabream (D2.6). In GMF and ARDAG, a single tank of bream broodstock were treated with an implant dose of 40 µg kg-1 GnRHa to induce spawning. Two spawns were collected before the application of the implant (control group) and two spawns after the application (hormone group). The tank in GMF had 22 females and 4 males of mixed wild and cultured origin to give a biomass of 82 kg that spawned 33 million floating eggs, which was approximately 0.4 million eggs.kg-1. In GMF, the hormone therapy approximately tripled egg production from 1 L per day to 3 L per day. The tank in ARDAG had 25 females and 9 males of cultured origin to give a biomass of 55kg that spawned 15 million floating eggs, which was approximately 0.3 million eggs.kg-1. In ARDAG, the hormone therapy maintained egg production constant when egg production appeared to be declining at the end of the spawning season. The parental contribution increased slightly in GMF, males remained similar before and after hormone application, with two males making 98 % of the contribution and female contribution increased from 9 and 10 females to 14 and 18 females participating. The number of families increased from 15 and 24 families before hormone application to 23 and 28 after hormone application. In ARDAG, a similar result was obtained, males remained similar before and after hormone application, with a total of seven to eight males participating both before and after and two males making 70 % of the contribution. However, the female contribution increased from 12 and 18 females to 19 and 21 females participating. The number of families increased from 33 and 45 families before hormone application to 44 and 50 after hormone application. The increase in families combined with the substantial increase in eggs spawned (GMF) increased the number of progeny available from different families for genetic selection.
In conclusion the guidelines of spawning protocols for optimised parental contribution (D2.7) were:
For European seabass:
1) set up large broodstock groups of 100+ mature fish;
2) at the peak of the spawning season, select mature broodstock based on careful examination of the stage of oocyte development;
3) induce with the female broodstock with 10 µg.kg-1 GnRHa 4) Collect eggs over two days of peak spawning.
For gilthead seabream:
1) set up large broodstock groups of 30+ mature fish;
2) at the peak of just past the peak of the spawning season induce all the broodstock with 40 µg.kg-1 implant of GnRHa;
3) Collect eggs over two days of spawning.
WP3
The aims of WP3 were to:
- screen available genetic sequences from seabass and seabream and further development for each species of two ultra-efficient and powerful multiplex sets (one with STRs and the other with SNPs) to be used for parentage assignment in captive broodstocks;
- assess spawning kinetics and manipulated hormonal spawning synchronisation in both species through the genetic analysis of full and half-sib groups.
Task 3.1. Development of a powerful STR 'gold standard' marker panel for each species
Associated deliverables:
D3.8.1: Screening of the STRs and SNPs public databases
D3.9: Development of a STRs multiplex for each species
In order to develop multiplexes for both species, two different public databases have been screened: for seabream and SIGENAE database (for seabass). Priority was given to loci proved to be highly polymorphic in previous studies and those located in known linkage groups.
Starting from some 100 loci in bream and 191 in bass we finally optimised one multiplex per species with 13 and 14 loci, respectively, for bream and bass.
Task 3.2. Development of a high resolution SNP parentage panel for each species
Associated deliverables:
D3.8.2: Screening of the STRs and SNPs public databases
D3.10: Development of a 60 SNPs microchip for each species
Data sets for European seabass (D. labrax) and Gilthead Seabream (S. aurata) were in the form of raw ACE assembling files for both Sparus aurata (ESTs assembling psb5 05/2010) and Dicentrachus labrax (ESTs assembling pdl5 05/2010) and were kindly provided by Cedric Cabau (see http://www.sigenae.org for details).
SNPs mining on the contigs was performed using three different softwares: GigaBayes (Marth et al. 1999), PolyFreq (Wang et al. 2005) and ssahaSNP (Ning et al. 2001) using default parameters settings. Results were inserted into a MySQL database for further filtering / sorting. For each SNP, the consensus ±80 bps flanking sequence was extracted and functional annotation of the ESTs was obtained from Sigenae. An additional BLASTN and BLASTX analyses with the stickleback (Gasterosteus aculeatus) from ENSEMBL (cDNA and pep version 59) was performed using NoBlast (Lagnel et al., 2009), and when a hit was reported the position on the stickleback genome was identified.
From approximately 49 000 and 37 000 ESTs for seabream and seabass, respectively, 488 and 405 SNPs were identified for the gilthead seabream and for the European seabass.
For both species, 128 SNPs were submitted for array synthesis at AB on July 2011 and arrived in both HCMR and ISILS by mid-November. These were chosen based on the number of softwares supporting them, the number of ESTs (SNPdepth), the conserved flanking region, the minor allele frequency, their position on the stickleback linkage map and their annotation. Trials started in December 2011.
Task 3.3. Use of multiplexes to describe spawning kinetics and manipulated hormonal spawning synchronisation in both species
Associated deliverables:
D3.11: Genotyping of broodstocks
D3.12: Genotyping of post-larvae and fry SME samples
D3.13: Genotyping of post-larvae samples (IRTA batches collected on a daily basis)
D3.14: Genotyping of post-larvae samples (IRTA batches obtained over two consecutive years).
Broodstock and larvae genotyping:
Using both STR (multiplex) and SNP (array) tools all available broodstocks were genotyped:
- 288 seabream from GMF Marine Farm S.A. (Greece), ARDAG Ltd (Israel), IRTA (Spain), Cupimar (Spain), TNSL (Spain),
- 235 seabass from GMF Marine Farm S.A. (Greece), IRTA (Spain), Cupimar (Spain) and TNSL (Spain).
In the seabream, we ended to 74 SNPs potentially useful in parentage assignment and population genetics studies: from the initial 128 SNPs ordered, 59 SNPs were polymorphic (3 expected clusters, score:
1) 10 SNPs did not show probably one class of homozygotes (less polymorphic SNPs with 2 clusters, score;
2) 38 SNPs presented close clusters (of which 5 SNPs were included in the 'selected' batch) and 21 SNPs failed to amplify correctly.
From this batch, 69 SNPs are considered to be of high quality for parentage assignment.
In the seabass, we ended to 78 SNPs potentially useful in parentage assignment and population genetics studies: from the initial 128 SNPs ordered, 44 SNPs were polymorphic (3 clusters), 34 SNPs were less reliable (either with close clusters or only two clusters), 32 SNPs close clusters (different levels) and 18 failed to amplify. The number of 'good quality SNPs' to be finally used in assignment tests could be reduced to 61 SNPs after the results that ISILS had obtained by duplicating some of the HCMR's experiments.
From the ISILS cross-checking, 24 SNPs were excluded in part because they were considered as monomorphic and in part as risky with close clusters; therefore, they were excluded from the initial 78 SNPs panel. But, it is also true, that ISILS screened these SNPs only in IRTA samples whereas HCMR had screened these SNPs on a larger number of broodstock and so, certainly, most of the SNPs that were considered as monomorphic for ISILS (and refer to only a subset of IRTA broodstock), are in reality polymorphic when taken in the totality of the broodstock samples. This is also the case of three SNPs excluded in the HCMR analysis and added in the one performed in ISILS since all showed two well-defined clusters; however, in the bigger dataset these are no good quality SNPs.
A cross species trial was performed checking the polymorphism of the seabream and seabass SNPs on seabass and seabream samples, respectively. For the seabream SNPs, 50 seabass samples were assayed. All the tested 74 SNPs showed clusters plots defined as 'failed', giving no possibility to enlarge the seabass panel with seabream SNPs. When results of the above mentioned 78 SNPs were checked on 23 wild seabream samples, 24 SNPs seemed to work pretty well and 13 SNPs showed lower quality results; all, or at least some of, these SNPs could be potentially added to a future new array in seabream.
The seabream microsatellite multiplex revealed to be highly efficient, with a rate of unambiguous allocations of 100 % for IRTA offspring (99.9 % of single match and 0.1 % of multiple match solved by stochastic approach) and 99.2 % for SME offspring (97 % of single match and 2.2 % of multiple match solved by stochastic approach). The remaining 0.8 % is represented by unsolved multiple assignments.
The seabass microsatellite multiplex gave contrasted results. The rate of unambiguous assignment reached 86.8 % in the IRTA offspring against 76.8 % for Cupimar and 50.3 % for TNSL. Looking at the tank level in the IRTA samples, we note that this rate rises to 95.3 % in V8 against 76.2 % in the V6 tank. The lower rates obtained in seabass can be explained in part by the presence of unsexed breeders in the farm broodstock and mixing of batches of different origins but also by possible genotyping scoring errors.
The 69 SNP panel tested on revealed to be very powerful, with 100 % of unambiguous allocation of which, 99,1 % solved in single match by deterministic approach and 0,9 % solved by stochastic approach (see deliverable 4.17 for more details). The lower rate obtained on seabass (71 %) can be interpreted by:
i) the presence of unsexed breeders in the seabass broodstock and erroneously mixed batches;
ii) the minor seabass DNA quality due to less performing equipment used for the DNA extraction on seabass (PrepStation versus BioSprint on seabream) and
iii) the supposed lower quality of the SNPs reads due to the use of the AutoLoader instead of the AccuFill machine, the latter being used to load the correct quantity of the DNA in the array.
Total number of seabream fish analysed with:
STRs: 4856
1608: SMEs offspring (802 ARDAG and 806 GMF) [4 batches (about 200 samples/batch)]
2960: IRTA offspring [daily and weekly spawning]
197: broodstock (112 ARDAG, 64 GMF, 24 IRTA).
SNPs: 514
226: SMEs offspring (ARDAG and GMF)
288: broodstock (ARDAG, GMF, IRTA, Cupimar, TNSL).
Total number of seabass fish analysed with:
STRs:3435
1634: SMEs offspring (802 ARDAG and 806 GMF) [4 batches (about 200 samples/batch)]
1566: IRTA offspring [daily and weekly spawning]
235: broodstock (116 TNSL, 30 IRTA, 39 CUPIMAR and 50 GMF).
SNPs: 859
481: SMEs offspring (312 Cupimar, 169 TNSL)
143: IRTA offspring [Daily and weekly spawning]
235: broodstock (116 TNSL, 30 IRTA, 39 Cupimar and 50 GMF).
WP4
The aims of WP4 were to:
- develop an allocation software working with STR and SNP markers to allow the accurate and efficient reconstruction of sibship with an easy-to-use procedure,
- perform the sibship reconstruction on every fry sample produced and compare the efficiency of the two types of markers (STRs and SNPs) to increase the rate of correct parentage allocation.
Tasks 4.1. and 4.2. New algorithms to solve multiple matches and produce virtual offspring batches
Associated deliverable:
D4.15: Software with simulation and parenthood procedures
A new version of the Windows-based software for parental allocation (PA) has been developed for the REPROSEL project. The software has been named parental allocation in diploid fish (PAF).
PAF uses the exclusion method as the main/first approach to allocate the parentage. In case of multiple allocations, it uses the likelihood approach. The identification of a parental pair at a given n-locus, with no previous information, is computed according to Marshall et al. (1998). For all the loci, the total likelihood (TL) is computed and, in case of multiple allocations, the difference between the two consecutive ranked TL values (?P) is computed. When ?P > 0 the parental allocation with the highest TL value is accepted and the multiple allocation is resolved. The program follows a stepwise procedure allowing the progressive exclusion of markers, which appears to be useful for SNPs due to their higher global risk of mismatch. Thanks to this procedure PAF distinguishes between the common ±2 pb errors and larger genotypic discrepancies.
The simulation-task, with the creation of Virtual Offspring, allows estimating the efficiency of the loci used in parentage assignment by the finding the ?P threshold that permits to obtain no more than 5 % of erroneous allocations (case of multiple allocations). A mating design is generated based on the parents' dataset. The percentage (or number) of males and/or females involved in the mating can be parameterised. The simulation-procedure produces a virtual offspring by random-sampling all the possible gamete combinations derived from the broodstock. This task can be used only for dataset with gender information on the broodstock.
Two input files are required, the parents and offspring files. They are organised in columns indicating the individual ID, marker ID, biallelic genotypes and sex for the broodfish, and can be sorted in whatever order. The missing values are automatically coded as 9 and SNPs scores (A, C, G, T) are converted in numbers (10, 20, 30, 40). The files are in dbf formats and can be easily generated by a simple copy / paste operation from any tab delimited text file. The call of the markers can be done by simply clicking on a button or pasted as text (markers are separated by a comma).
The main outputs are the parental allocations with their probability estimation, the crossing scheme, the observed and expected heterozygosities and the PIC and exclusion probabilities. They are presented in txt or dbf formats but can also be produced in other formats when requested.
Tasks 4.3. and 4.4. Reconstruction of sibship and cost-effectiveness of the marker panels
Associated deliverables:
D4.16: PA obtained on all fry samples
D4.17: Cost-effectiveness of SNPs and STRs panels for PA.
PA (additional information under WP3):
The PAs were obtained with PAF using the two allocation methods (exclusion and probabilistic). Large genotypic discrepancies were handled accepting a maximum of 2 mismatches for the STRs panels and 3 mismatches for the SNPs panels, while allowing for any ±2 pb errors.
The majority of the offspring was allocated with the STRs panels (13 and 14 markers for seabream and seabass, respectively) while a minor part derived from the SNP validation procedure was obtained with the SNPs panels (69 and 78 markers for seabream and seabass, respectively). Some undetermined broodfish could be sexed and some offspring batches that had been erroneously mixed were evidenced thanks to the allocation results.
Cost-effectiveness:
The allocation efficiency has been tested looking at the effect of increasing the number of breeders (up to 276) to allocate a fixed number of seabream offspring belonging to known families (107 from ARDAG and 119 from GMF). Due to biases in the seabass batches (broodfish incompletely sexed and offspring lots erroneously mixed), the case study has been limited to the seabream batches only. Breeders having more than 15 un-scored SNPs were preventively eliminated (12 in total) in order to avoid biased multiple allocations. This precaution needs to be pointed out for SNPs as the absence of genotype is interpreted by the allocation software as a valid match and thereby increases the percentage of false assignments.
When breeders from the original experimental tanks are considered, the 69 SNPs panel appears to outperform the 13 STRs multiplex by assigning all the true parents in a single match and by solving STRs unassigned multiple matches. Instead, the use of larger (fictive) broodstock increases the rate of erroneous assignments for both types of markers, in particular with SNPs when true and putative parents have similar genotypes. However, when breeders share a limited number of common alleles, SNPs are more efficient than STRs to find the correct parental pairs, allowing in the case of the ARDAG batch a 100 % unambiguous allocation when all the 276 fictive breeders are considered.
The informativeness of the seabass and seabream markers panels were calculated with the polymorphic information content (or PIC). The low PIC values of the SNPs, due to their low allelic diversity, is compensated by non-exclusion probability values similar to STRs for seabream, higher for seabass but still close to zero, indicating a high capacity of both marker panels to exclude false random parents. Removing markers in the panels (two STRs or four SNPs) did not change significantly these values.
When comparing the typing with the two types of markers, we note that STRs often becomes tedious and time consuming, and represents a real bottleneck when high number of samples is processed. In the case of SNPs, the sample processing may be completely automated and the genotype can be obtained rapidly without requiring manual checks. At present, at least four manufacturers propose genotyping solutions adapted to parentage allocation: Illumina, Sequenom, Life Technologies (Applied Biosystems) and Fluidigm. They propose flexible array combinations to choose the genotyping platform adapted to every specific need (how many markers on how many samples).
WP5
The aims of WP5 were to optimise the design of single and multiple trait mass spawning selective breeding schemes taking into account the reproductive features of seabass and seabream populations of the SMEs.
Associated deliverables:
D5.18: Optimised mass spawning breeding scheme designs for commercial broodstocks
D5.19: Guidelines on breeding schemes implementation.
In collabouration with each of the SME partners, information from WPs 2 to 4 on the achieved contributions of males and females in a typical mass spawning tank in their facility for European seabass or gilthead seabream has been used to give an advice on a design of a breeding program. Computer simulations of their specific designs have been done for this optimisation step, taking into account of the reproductive limitations of their respective species. This work was done in close collabouration between the SMEs and the R&D partner Nofima. The result has been delivered in the forms of reports containing information about the selection procedure, the trait to select for, how many tanks needed and other information specific to each of the SMEs. In addition, Nofima has provided the SMEs with a software to select males and females while controlling the rate of inbreeding to the desired rate of 1 %, which can also be used to choose which parents to mate.
A guideline on practical seabass and seabream breeding schemes has been written. The aim of the guideline is to provide the SMEs with general recommendations of the different steps required for setting up a practical mass spawning breeding scheme for seabass and seabream. When differences between species are important for the breeding design, their specific characteristics have been taken into account. Differences in production systems and breeding stock between the SMEs have been considered in the specific SME reports of deliverable 5.18.
The guideline first explains the basis of selective breeding theory, which is an important prerequisite for understanding the guidelines for the mass spawning breeding scheme. Secondly, the guidelines explain how to set up a base population, which should contain high genetic variation, because this population serves as the basis for all coming generations. SMEs are provided with guidelines on how to choose their base population.
Breeding designs are optimised by comparing obtained rate of inbreeding and genetic gain using a stochastic simulation model of the breeding program. In the simulations the range of each parameter on input is from participating SMEs and results obtained in the REPROSEL WPs2 to 4. The focus is on how to obtain highest possible genetic gain per generation when at the same time keeping rate of inbreeding at a low level. In the basic mass spawning breeding scheme, selection of male and female brood fish is done in two steps. There are 50 selected sires and 100 selected dams, which are distributed into 5 mass spawning tanks. These sires and dams are mated randomly, all have different contributions and are producing a total of 10 000 offspring in 5 different mass spawning tanks. After hatching and the grow-out period, 250 males and 250 females are preselected out of the 10 000 fry based purely on their phenotypic weight record. These 500 preselected fish then need to be genotyped in order to assign them to their parents. When the parentage assignment is done the fish are added to the pedigree and BLUP breeding values are calculated based on records of weight of the fish, and relationship within the current generation as well as other generations of fish. From these preselected fish, the best dams and sires with respect to breeding values are selected. These selected sires and dams then go into the mass spawning tanks and become potential parents of the next generation. Due to the different generation intervals, females and males used in each year do not belong to the same year class. This will have a positive effect of the rate of inbreeding per year.
Results show that with respect to rate of inbreeding and genetic gain, 500 to 1000 genotyped offspring seems to be optimal. However, this needs also to be compared with the cost of genotyping. Results also show that based on the contribution of males and females applied in this study, at least 5 different tanks are needed to in order to keep the rate of inbreeding below the general accepted limit of 1 %. A tool to control inbreeding in a practical mass spawning breeding scheme is also presented and has been provided to the SMEs. The tool puts a cost-factor on inbreeding to the estimated breeding values, such that parents are selected based on their relationship in order to get a rate of inbreeding below 1 % per year. In the guidelines it is also discussed how to be able to select for multiple traits. By applying the suggested breeding schemes, it is possible to achieve a genetic gain in growth of at least 10 % per generation.
Potential impact:
In a general way the role played by the genetic markers in selective breeding programs is constantly increasing. In REPROSEL, these markers are first intended to rebuild the pedigree of fish originated from different families and mixed from the egg stage. But such markers need to be cost-effective to change the breeding practices and be more widely used in commercial programs. We believe that the SNP panels developed during the project have the requisites to be effective for parentage inference and also to lower the genotyping costs. The current and near future high throughput genotyping technologies should make the difference and limit the cost to some Euros per individual (today the floor price is inferior to EUR 0.07/SNP).
Cost-effective markers represent as well a strong incentive to implement a breeding program as they permit the rearing tanks used for production to be easily converted for the convenience of the program, so avoiding costly investments for specific facilities and associated high running costs.
The current reproduction protocols suffer from knowledge gaps on mass spawning control. The results of WP2 show how strongly biased the family structure of a traditional offspring batch can be for both studied species. The mass spawning protocols and guidelines (D2.6 and D2.7) explain how to take into account of the biasing factors (maturation stage, sex ratio, number of breeders, tank size, egg collection) and produce the breeding population with optimised parental contribution.
The allocation software PAF has been widely tested on the progeny batches produced during the project (over 8000 offspring from 8 different broodstock) while a part of the assignment results has been compared with other allocation programs (CERVUS and VITASSIGN) to validate the algorithms. The successive versions of the program have finally produced an efficient and easy-to-use tool working both on STR and SNP markers. The scalability of the outputs allows a smart interpretation of the results, starting from the turnkey mating scheme up to key information such as marker incompatibilities and allocation power. In synthesis, PAF is designed for a direct use in the industry and is perfectly adapted to SNPs, presented as valid alternative to STRs.
The guidelines on breeding schemes suggest preselecting a subgroup of the selection candidates of 500 - 2000 fish and, when possible, choose groups of 100 to 200 fish at the time for genotyping. This procedure permits to limit the cost of the program by containing the costly parentage assignment expenses. Having the pedigree information allows calculating the breeding value of the candidates and so getting higher accuracy in selecting the heritable component of the trait under improvement. The breeding scheme describes also a method to keep the rate of inbreeding around 1 % by applying a cost factor parameter when calculating the breeding value. This cost factor can be easily set by using the executable tool provided to the SME partners.
The project results as a whole will contribute to the implementation of sustainable breeding programs. Having the tools to obtain the pedigree information and to increase the effective population size, the SMEs will improve their broodstock while keeping the rate of inbreeding under control. They will increase their competitiveness on the fry market where the share of selected animals is growing continuously.
The foreground on mass spawning will serve the breeding program but will also be useful in production to improve the broodstock management in the hatchery and the fry throughput in the nursery.
Dissemination activities
Communication strategy
The plan to reach the target public has been initiated following the statements of the consortium agreement regarding the confidentiality of the foreground and the publication rights arising from it. The main undertaking was not to jeopardise the immediate and exclusive exploitation of the results by the participating SMEs.
During the course of the project, further discussions showed a divergence between the academic achievement criteria and the industry expectations regarding the dissemination strategy. To help define a common approach, the four participating SMEs were asked to express their preferences on the way each specific foreground should be managed. The questions related to the project foreground were:
I: Do you consider that the information should not be released for now because it may put at risk the exploitation plans?
II: Do you consider that the information can be published outside the consortium now (scientific journals, press release, website public area...)? prior viewing by the SMEs is needed?
III: Do you consider that the information is worth to be licensed to third party upon specific agreement?
IV: Other choice if the above propositions do not suit you?
Any release of non-confidential information outside the consortium was done under the condition of the prior consent of the SME representatives. The release of this information has been done through two main channels:
A. Project website (see http://www.reprosel.eu online)
The REPROSEL website has a public area dedicated to the dissemination of the project as a whole. It gives a general presentation of the project partnership, objectives and main workpackages. The 'Results' section allows reaching a large public with the downloadable abstract of the main project results, while avoiding to give sensitive details to preserve the SMEs interest.
B. Congresses and conferences
To disseminate towards the industry and the scientific community, the REPROSEL results were presented in various European conferences between May and September 2012.
Dissemination actions
Three European conferences have been chosen to present the results of the project:
- 1st Balkan Conference on the Biology of Reproduction in Farm Animals and in Aquaculture, Palace of Culture, Tirana, Albania, 23 - 25 May 2012,
- 2012 Annual Symposium of the Fisheries Society of the British Isles, University of East Anglia, Norwich, UK, 9 - 13 July 2012,
- Aqua 2012 Global aquaculture - Securing our future, Prague, Czech Republic, 1 - 5 September 2012.
In all, four oral presentations and one poster have been presented.
Although not part of the REPROSEL expected results, the work about the spawning behaviour of gilthead seabream derives from the REPROSEL broodfish tanks of IRTA and provides useful information to depict the reproductive behaviour of the species.
The presentations (and their abstracts) can be consulted in the 'Results' webpage (see http://www.reprosel.eu/SitePages/Results.aspx online).
Two other publications referring to the approach explored in REPROSEL have been written in 2012. One is a book chapter to be published in 2013, the other work is already published in a peer-reviewed periodical.
The peer-review publications have been limited not to disclose sensitive information that may jeopardise the future exploitability of the results. However, the project findings will allow in the near future the RTD partners to disseminate more widely the results in scientific journals.
In a more general way internet has also been used to advertise the project, as have done some partners on their website:
Nofima:
http://www.nofima.no/en/prosjekt/reprosel
IRTA:
http://www.irta.cat/en-us/rit/projectes/pages/ProjectDisplayPage.aspx?UrlCode=844
ISILS:
http://www.istitutospallanzani.it/scheda.php?id=133
Exploitation of the results
Ownership of the project results
The foreground of the project is the sole and exclusive property of the SMEs. The joint ownership has been chosen for all the results except for one confidential deliverable (D5.18) specific to each SME.
Presentation of the results
An important step before the delivery of the project results to the SMEs has been the training sessions organised by the RTD partners during the last project meeting. The SME partners took actively part to the sessions, involving also other technical and managerial personnel of the companies. For each of the four workpackages related to RTD activities (WPs 2 to 5), half a day has been dedicated to the presentation of the results followed by an open discussion on the way they should be applied. Demonstrations have been made on how to use the allocation software and how to estimate essential parameters in a breeding programme. The ppt presentations are available in the private REPROSEL webspace at 'WP / Meetings / Final meeting Heraklion'.
A person external to the REPROSEL consortium (Cameron Brown from the Cyprus University of Technology) having done pioneering work on seabream parental contribution, has presented unpublished results of interest for the partners.
As most of the results were not fully completed at the last meeting or needed to be improved further to specific requests, the definitive results have been delivered to the SMEs as a unique package later in December.
How the foreground will be exploited
The advancement level of the genetic program is not the same in all the SMEs, which explains the different approaches and timelines to use the results. Practically, with the project results in hands the SMEs will be able to define their own strategy regarding:
- the base population to start the breeding programme,
- the number of breeders and the sex-ratio forming the breeding nucleus,
- the number and size of the tanks holding the breeders,
- the broodstock rearing protocol (including hormonal treatment) to be used for mass spawning and optimised family structure,
- the size, number and frequency of the progeny batches to be sampled,
- the way each batch should be reared (mixed vs separate) before tagging,
- the selected trait and the recording age / size (sex differentiation, precocious selection),
- the number of fish to be measured, PIT-tagged and fin-clipped,
- the markers panels to be used, the genotyping platform and, associated to it, the genotyping laboratory,
- the way the fish will be genotyped, at once or in a stepwise way,
- the way each candidate will be selected, based on its phenotype or its breeding value (adjusted or not according to their parentage relationship),
- the number of selected males and females and the associated selection intensity,
- the way the new broodfish tanks will be constituted,
- the way each generation will be produced (interval and overlapping).
The guidelines of spawning protocols (D2.6 and D2.7) and breeding schemes (D5.19) indicate for each species how to set up the broodstock groups and collect the eggs to optimise the parental contribution. The screened markers (D3.9 and D3.10) and cost-effectiveness study (D4.17) lists the best performing panels and the current and future genotyping solutions to get the pedigree of the selection candidates. A list of biotech companies providing a genotyping service will help SMEs to choose the best laboratory adapted to their needs. The PAF allocation software (4.15) has been developed for kinship reconstruction with both types of markers; it presents an easy-to-use interface and provides useful outputs for the industry. The cost-factor executable (D5.19) allows the SMEs to calculate the breeding value of the candidates while accounting for the inbreeding level. Finally, a confidential document describes the main steps to implement a breeding scheme adapted to the specificities of each SME (D5.18).
At present no IPR measure has been taken by any of the SMEs, although an interest exists to exploit the marker panels and the allocation software through licensing.
Project website address:
http://www.reprosel.eu/
Contact details:
Nofima:
Herve Chavanne
E-mail: herve.chavanne@nofima.no
Anne Risbrathe
E-mail: anne.risbraathe@nofima.no
Telephone: +47-649-70326
Anna Kristina Sonesson
E-mail: anna.sonesson@nofima.no
Telephone: +47-649-40849
HCMR:
Costas Tsigenopoulos
E-mail: tsigeno@her.hcmr.gr
Telephone: +30-281-0337854
IRTA:
Neil Duncan
E-mail: Neil.Duncan@irta.cat
Telephone: +34-977-745427
ISILS:
Katia Parati
E-mail: katia.parati@istitutospallanzani.it
Telephone: +39-036-378883
GMF
Nikolaos Papaioannou
E-mail: papaioannou@irida-sa.gr
Telephone: +30-697-8182023
ARDAG
Glen Pagelson
E-mail: glen@ardag.co.il
Telephone: +97-286-335111
Cupimar:
Alfonso Vidaurreta
E-mail: campillo@cica.es
Telephone: +34-956-883447
TNSL:
Carlos Mazorra
E-mail: carlosmazorra@tinamenor.es
Telephone: +34-942-718020,
We describe the daily and weekly spawning kinetics of cultured broodstock analysing the family structure of single spawning events (45 for the two species) over the entire reproduction period. A further analysis of the parental contribution of hormonally induced commercial broodstock (16 batches of 200 offspring) allowed developing protocols and guidelines on mass spawning synchronisation. New relative to the size of the broodstock appeared to be affected by the hormonal induction but also by the sex ratio, the number of breeders and the size of the tank.
Public databases were screened to develop ultra-efficient microsatellite (STR) and single nucleotide polymorphism (SNP) panels for parentage assignment. 13 seabream and 14 seabass STRs mapped in different linkage groups were selected to be amplifiable in a single tube. Each multiplex, first tested on known families, was validated on the farms' breeders (288 seabream and 235 seabass). SNPs were mined from the Sigenae database. They were detected based on the number of software supporting them with constraints on depth, minimum allele frequency, flanking sequences and GC%, and further annotated and positioned on the stickleback linkage map with ENSEMBL. From more than 400 identified SNPs per species, 128 were submitted for array synthesis (AB). 69 seabream and 61 seabass SNPs were validated as highly efficient for parentage assignment. We show that the assignment power of a large STR multiplex can be outperformed by an optimised SNPs panel having five times more markers.
A new allocation software was developed. It uses the exclusion method as the first approach to allocate the parentage. In case of multiple allocations, it uses the likelihood approach. The program follows a stepwise procedure allowing the progressive exclusion of markers, which appears to be useful for SNPs due to their higher global risk of mismatch. Common ±2 pb errors and larger genotypic discrepancies remain distinguishable. Two input files, parents and offspring, are required and can be easily generated by a simple copy / paste operation from any tab delimited text file. The main outputs are the parental allocations with their probability estimation, the crossing scheme, the observed and expected heterozygosities and the PIC and exclusion probabilities.
Breeding designs were optimised by comparing obtained rate of inbreeding and genetic gain using a stochastic simulation model of the breeding program. The simulations took into consideration the number, sex-ratio, random mating and contribution of breeders and the number of broodstock tanks. A stepwise approach is proposed to genotype and select groups of fish based on their pedigree and BLUP breeding values, so contributing to limit the genotyping costs. A tool that puts a cost-factor on inbreeding to the estimated breeding values has been developed to control the rate of inbreeding. By applying the suggested breeding schemes, it is possible to achieve a genetic gain in growth of at least 10 % per generation.
Project context and objectives:
Selective breeding programs require long-term investment, and their results depend on the compromise between the genetic response obtained and the maintenance of the genetic variability, i.e. controlling rates of inbreeding. These two fundamentals may be put at risk when choosing separate rearing of full and half-sib groups produced by different parents, in situations where:
(a) significant between-tank effects increase the phenotypic variability affecting the genetic value estimates of sib groups; and
(b) lack of reproduction control results in a reduction in the number of families produced at each generation and skewed contributions of parents, resulting in increased rates of inbreeding.
The above risks are especially high for species characterised by batch-spawning strategies. Moreover, the man-power, facilities and broodstock required for an effective breeding program based on separate family rearing is a strong discouraging factor for SMEs to commit to family-based breeding programmes.
In salmonids, breeding schemes are based upon the manual stripping of mature gametes from the chosen broodfish and the artificial fertilisation of these gametes. For other species like the European seabass and the Gilthead seabream, characterised by batch-spawning strategies, the outcome of artificial fertilisation is awkward. The main limitations in the females derive from:
(a) the difficulty to synchronize the ovulation and spawning of numerous females in a single spawning event;
(b) the failure to predict accurately the time of ovulation in order to obtain the eggs at a prime condition;
(c) the very brief period after ovulation (1 -3 hours) before over-ripening can reduce egg viability to 0 %; and
(d) the limited quantity of good quality eggs available from one mature female in the case of the Gilthead seabream.
In these conditions, the optimisation of mass spawning to produce adequate pools of families and the development of competitive molecular panels for parentage assignment may become a valid alternative to implement selective breeding programs in these species. The main objective of REPROSEL is to provide the participating SMEs with tools to implement breeding programs with constrained rates of inbreeding, but without the necessity of utilising artificial insemination.
Batches of progenies will be obtained first from mass spawning of commercial broodstock raised in traditional production facilities to depict the parental contribution to the new generation, an opportunity which will open new strategies for broodstock management. Other progeny batches will be produced in experimental facilities using smaller broodstock tanks in order to better understand the spawning kinetics over the whole production season. Relying on mass spawning will eliminate the stress and potential damage of broodfish ensuing from the stripping operations and will avoid the disruption of production routine and the financial costs that it entails.
STR panels are already available for parentage inference on both species but most often their power does not bring to the high level of parentage assignment expected in commercial selective programmes. Consequently, new highly discriminative molecular panels for parentage assignment will be developed using STR and SNP markers, the latter being seen as a valid alternative to STRs. Their efficiency and costs will be compared.
The efficient sibship reconstruction, allowing a high and reliable level of unambiguous parentage assignment, is essential to build-up on the higher power of the markers panels. A new allocation software, combining the exclusion and probabilistic approaches, will be developed and adapted to the specificity of SNP markers. The objective is also to deliver an easy-to-use tool to prepare the input files and interpret the output information.
Families derived from a mass spawning event will keep maintaining an unfavourable structure for the needs of selective breeding, even if improvements are expected from the better understanding of the reproduction patterns of the species. Stochastic simulations will be used to account for such limits and propose breeding schemes adapted to the participating SMEs. Lack of identification of individuals and skewed contributions of parents will be compensated by an increased total number of individuals in order to assure wanted family sizes and increased number of batches. The different selection schemes to be explored will have controlled rates of inbreeding. These are two factors which reduce the risk of the breeding program, and together with multi-trait selection assure sustainable breeding programs for the SMEs.
With the proposed approach, the separate rearing of sibling groups will be replaced by the rearing in one or few communal tanks. The possibility to start the breeding evaluation using one or few tanks constitutes a considerable progress in term of required facilities and associated costs, and this will facilitate the access of SMEs to the proven benefits of selective breeding programmes. This approach is also expected to reduce undesired initial tank effects and the risks edging on the genetic evaluation inherent to the confusion between genetic and non-genetic effects.
Project results:
WP2
The aims of WP2 were to:
- produce post-larvae and fry batches from captive commercial and experimental seabass and seabream broodstock,
- describe the spawning kinetics and parental contribution of captive broodstocks and increase parental contribution in commercial hatcheries using hormonal manipulations to synchronize spawning,
- develop and propose a protocol for the increase of the number of spawning parents and equalise parental contribution, to be adopted by the industry.
Task 2.1. Description of spawning kinetics and parental contribution
Associated deliverables:
D2.2: Biological sampling of broodstock
D2.3: Synchronisation of spawning and production of fry batches (year 1)
D2.4: Synchronisation of spawning and production of fry batches (year 2)
D2.5: Description of spawning kinetics and endocrine control of seabass and seabream broodstock
D2.6: Development of spawning protocols for optimised parental contribution
D2.7: Guidelines of spawning protocols for optimised parental contribution.
IRTA set up and produced the required larval and blood samples from three tanks of European seabass broodstock and two tanks of gilthead seabream. Each of the four SMEs set up broodstock groups that were induced with the identified hormone protocols to produce the required seabass (TNSL and CUPIMAR) and seabream (ARDAG and GMF) larval samples. Broodstock fin-clips, larval and fry samples were collected.
The European seabass spawning kinetics were described (D2.5) in two tanks of cultured broodstock held in IRTA. Each tank had a biomass of 28 kg of seabass, tank A had five males and five females and tank B had four females and six males. Tanks A and B produced respectively, 11 and 15 spawns with approximate fecundities of floating eggs over the entire spawning season of 75 000 eggs.kg-1 and 100 500 eggs.kg-1. Both tanks started the spawning season with a single natural spontaneous spawn from a single female fertilised by three males. After this initial spawn the females were induced to spawn every 10 days with a single injection of 10 µg.kg-1 GnRHa (gonadotropin releasing hormone agonist), whilst males were implanted once with a slow release implant, 20 µg.kg-1. The hormone inductions produced four spawning periods in each tank, one induction in tank B produced three spawns, five inductions produced two spawns and two inductions, the last successful spawning in each tank was a single spawn. Both tanks had dominant males that participated in most spawns analysed, in tank A one male fertilised 73 % (participating in all spawns) of the larvae analysed and in tank B, two males fertilised 43 % and 47 % of the larvae analysed. All other males fertilised 0 % to 12 % of the larvae. Female contributions were also unequal, in tank A, two females contributed 45 % and 42 % of the larvae, one female 13 % and two females did not contribute. In tank B, one female contributed 66 %, two females 18 % and 16 % and one female did not contribute. One female in tank B contributed to all spawning periods, the initial spontaneous spawn and all four inductions to give a total of five spawns. Generally all spawns were from a single female (69 % of spawns) fertilised by three males (46 %) although the participation of just one (15 %), two (15 %) and four (15 %) males was also observed. The mean number of families obtained from a spawn was 3.2 ± 1.4 families and the mean number of families obtained from a hormone induction (usually two spawns) was 5.3 ± 1.2 families.
Plasma profiles of hormone changes (D2.5) exhibited clear peaks in estradiol (3.71 ± 0.83 ng.mL-1) testosterone (6.26 ± 2.61 ng.mL-1) and 11-ketotestosterone (2.67 ± 0.38 ng.mL-1) during 2012, when seabass were observed to spawn. No peaks and approximately half the plasma hormone level were observed during 2011, which presented poor spawning. Levels in wild and cultured broodstock were similar indicating that any reduction in spawning in cultured stocks was not due to significant decreases in hormone profiles.
The gilthead seabream spawning kinetics were described (D2.5) in two tanks of cultured broodstock held in IRTA. Each tank had a biomass of 34 - 35 kg of seabream, which consisted of 5 males and 7 females. Both tanks produced daily natural spontaneous spawning, tanks C1 and C2 produced respectively 121 and 120 spawns with approximate fecundities of floating eggs over the entire spawning season of 1.2 million eggs.kg-1 and 0.9 million eggs.kg-1. The spawning in both tanks lasted over four months from late December / early January to late May. Spawns were collected and analysed at approximately weekly intervals during the first two months of spawning and the five to six days of consecutive daily spawning were analysed from the fifth week. Both tanks had dominant males that participated in all spawns analysed, in C1 one male fertilised 91 % of the larvae analysed and in tank C1, one male fertilised 67 % and a second male 25 % of the larvae analysed. Both tanks also contained a male that did not participate in any spawning. Female contributions were more even and in C1, the contribution of four females varied from 15 % to 37 %, while three females contributed from 0 % to 2 %. In tank C2, the contribution of three females varied from 24 % to 46 %, while four females contributed from 0 % to 2 %. During the period of consecutive days of spawning, both females and males were observed to spawn daily. In tank C1, two females and one male spawned every day during five days and tank C2, one male spawned every day during six days and another male and two females spawned on five of the six days. Generally most spawns were contributed to by two males and two females and the mean number of families obtained from a spawn was 5.6 ± 2.5 families. There was a tendency for the number of families to increase from the early spawns in the season to the last spawns analysed that coincided with the end of the second month and the start of the third month of the 4+ month spawning season.
Task 2.2. and 2.3: Synchronisation of spawning in European seabass and Gilthead seabream
Associated deliverables: D2.2 D2.3 D2.4 D2.6 D2.7
Spawning protocols were developed to optimise the parental contribution of spawning European seabass (D2.6). In Cupimar, two tanks of wild seabass were used, one tank was a control tank from which natural spontaneous spawns were collected and the second tank was hormonally induced, the females were induced to spawn with a single injection of 10 µg.kg-1 GnRHa, whilst males were implanted once with a slow release implant, 20 µg.kg-1 GnRH. The control tank had 13 males and six females with a total biomass of 51 kg. The hormone tank had nine males and nine females and two unidentified fish with a total biomass of 56 kg. The volume of eggs obtained from the two tanks was similar, the control gave 1.6 L of floating eggs, which was approximately 23 250 eggs.kg-1 and the hormone group gave 1.4 L of floating eggs, which was approximately 19 500 eggs.kg-1. In the control group, four females and seven males participated to give 15 - 17 families. The participation of individuals increased slightly in the hormone group and six females and seven males participated, but the number of families (13 - 15) was slightly lower. In TNSL, the same large group of 116 cultured seabass (size tank) was hormonally induced to spawn over consecutive years. The group consisted of 45 females, 62 males and 9 unidentified fish that had a biomass in 2011 of 232 kg and in 2012 of 330 kg. In 2011, just the females were induced using a single injection of 10 µg.kg-1 GnRHa, whilst in 2012 the females were given the same hormonal induction and the males were implanted once with a slow release implant, 20 µg.kg-1 GnRHa. In 2011, 13.7 million floating eggs were collected, which represented approximately 58 900 eggs.kg-1 and in 2012 8 million floating eggs were collected, which was approximately 24 300 eggs.kg-1. Two spawns were analysed from each year (four spawns in total) and the parental contribution was similar with a mean participation of 24 ± 8 males and of 17 ± 2 females to produce a mean of 41 ± 10 families. The large group of broodstock combined with hormone induction appeared to increase the number of families compared to the smaller groups in Cupimar and IRTA.
Spawning protocols were developed to optimise the parental contribution of spawning gilthead seabream (D2.6). In GMF and ARDAG, a single tank of bream broodstock were treated with an implant dose of 40 µg kg-1 GnRHa to induce spawning. Two spawns were collected before the application of the implant (control group) and two spawns after the application (hormone group). The tank in GMF had 22 females and 4 males of mixed wild and cultured origin to give a biomass of 82 kg that spawned 33 million floating eggs, which was approximately 0.4 million eggs.kg-1. In GMF, the hormone therapy approximately tripled egg production from 1 L per day to 3 L per day. The tank in ARDAG had 25 females and 9 males of cultured origin to give a biomass of 55kg that spawned 15 million floating eggs, which was approximately 0.3 million eggs.kg-1. In ARDAG, the hormone therapy maintained egg production constant when egg production appeared to be declining at the end of the spawning season. The parental contribution increased slightly in GMF, males remained similar before and after hormone application, with two males making 98 % of the contribution and female contribution increased from 9 and 10 females to 14 and 18 females participating. The number of families increased from 15 and 24 families before hormone application to 23 and 28 after hormone application. In ARDAG, a similar result was obtained, males remained similar before and after hormone application, with a total of seven to eight males participating both before and after and two males making 70 % of the contribution. However, the female contribution increased from 12 and 18 females to 19 and 21 females participating. The number of families increased from 33 and 45 families before hormone application to 44 and 50 after hormone application. The increase in families combined with the substantial increase in eggs spawned (GMF) increased the number of progeny available from different families for genetic selection.
In conclusion the guidelines of spawning protocols for optimised parental contribution (D2.7) were:
For European seabass:
1) set up large broodstock groups of 100+ mature fish;
2) at the peak of the spawning season, select mature broodstock based on careful examination of the stage of oocyte development;
3) induce with the female broodstock with 10 µg.kg-1 GnRHa 4) Collect eggs over two days of peak spawning.
For gilthead seabream:
1) set up large broodstock groups of 30+ mature fish;
2) at the peak of just past the peak of the spawning season induce all the broodstock with 40 µg.kg-1 implant of GnRHa;
3) Collect eggs over two days of spawning.
WP3
The aims of WP3 were to:
- screen available genetic sequences from seabass and seabream and further development for each species of two ultra-efficient and powerful multiplex sets (one with STRs and the other with SNPs) to be used for parentage assignment in captive broodstocks;
- assess spawning kinetics and manipulated hormonal spawning synchronisation in both species through the genetic analysis of full and half-sib groups.
Task 3.1. Development of a powerful STR 'gold standard' marker panel for each species
Associated deliverables:
D3.8.1: Screening of the STRs and SNPs public databases
D3.9: Development of a STRs multiplex for each species
In order to develop multiplexes for both species, two different public databases have been screened: for seabream and SIGENAE database (for seabass). Priority was given to loci proved to be highly polymorphic in previous studies and those located in known linkage groups.
Starting from some 100 loci in bream and 191 in bass we finally optimised one multiplex per species with 13 and 14 loci, respectively, for bream and bass.
Task 3.2. Development of a high resolution SNP parentage panel for each species
Associated deliverables:
D3.8.2: Screening of the STRs and SNPs public databases
D3.10: Development of a 60 SNPs microchip for each species
Data sets for European seabass (D. labrax) and Gilthead Seabream (S. aurata) were in the form of raw ACE assembling files for both Sparus aurata (ESTs assembling psb5 05/2010) and Dicentrachus labrax (ESTs assembling pdl5 05/2010) and were kindly provided by Cedric Cabau (see http://www.sigenae.org for details).
SNPs mining on the contigs was performed using three different softwares: GigaBayes (Marth et al. 1999), PolyFreq (Wang et al. 2005) and ssahaSNP (Ning et al. 2001) using default parameters settings. Results were inserted into a MySQL database for further filtering / sorting. For each SNP, the consensus ±80 bps flanking sequence was extracted and functional annotation of the ESTs was obtained from Sigenae. An additional BLASTN and BLASTX analyses with the stickleback (Gasterosteus aculeatus) from ENSEMBL (cDNA and pep version 59) was performed using NoBlast (Lagnel et al., 2009), and when a hit was reported the position on the stickleback genome was identified.
From approximately 49 000 and 37 000 ESTs for seabream and seabass, respectively, 488 and 405 SNPs were identified for the gilthead seabream and for the European seabass.
For both species, 128 SNPs were submitted for array synthesis at AB on July 2011 and arrived in both HCMR and ISILS by mid-November. These were chosen based on the number of softwares supporting them, the number of ESTs (SNPdepth), the conserved flanking region, the minor allele frequency, their position on the stickleback linkage map and their annotation. Trials started in December 2011.
Task 3.3. Use of multiplexes to describe spawning kinetics and manipulated hormonal spawning synchronisation in both species
Associated deliverables:
D3.11: Genotyping of broodstocks
D3.12: Genotyping of post-larvae and fry SME samples
D3.13: Genotyping of post-larvae samples (IRTA batches collected on a daily basis)
D3.14: Genotyping of post-larvae samples (IRTA batches obtained over two consecutive years).
Broodstock and larvae genotyping:
Using both STR (multiplex) and SNP (array) tools all available broodstocks were genotyped:
- 288 seabream from GMF Marine Farm S.A. (Greece), ARDAG Ltd (Israel), IRTA (Spain), Cupimar (Spain), TNSL (Spain),
- 235 seabass from GMF Marine Farm S.A. (Greece), IRTA (Spain), Cupimar (Spain) and TNSL (Spain).
In the seabream, we ended to 74 SNPs potentially useful in parentage assignment and population genetics studies: from the initial 128 SNPs ordered, 59 SNPs were polymorphic (3 expected clusters, score:
1) 10 SNPs did not show probably one class of homozygotes (less polymorphic SNPs with 2 clusters, score;
2) 38 SNPs presented close clusters (of which 5 SNPs were included in the 'selected' batch) and 21 SNPs failed to amplify correctly.
From this batch, 69 SNPs are considered to be of high quality for parentage assignment.
In the seabass, we ended to 78 SNPs potentially useful in parentage assignment and population genetics studies: from the initial 128 SNPs ordered, 44 SNPs were polymorphic (3 clusters), 34 SNPs were less reliable (either with close clusters or only two clusters), 32 SNPs close clusters (different levels) and 18 failed to amplify. The number of 'good quality SNPs' to be finally used in assignment tests could be reduced to 61 SNPs after the results that ISILS had obtained by duplicating some of the HCMR's experiments.
From the ISILS cross-checking, 24 SNPs were excluded in part because they were considered as monomorphic and in part as risky with close clusters; therefore, they were excluded from the initial 78 SNPs panel. But, it is also true, that ISILS screened these SNPs only in IRTA samples whereas HCMR had screened these SNPs on a larger number of broodstock and so, certainly, most of the SNPs that were considered as monomorphic for ISILS (and refer to only a subset of IRTA broodstock), are in reality polymorphic when taken in the totality of the broodstock samples. This is also the case of three SNPs excluded in the HCMR analysis and added in the one performed in ISILS since all showed two well-defined clusters; however, in the bigger dataset these are no good quality SNPs.
A cross species trial was performed checking the polymorphism of the seabream and seabass SNPs on seabass and seabream samples, respectively. For the seabream SNPs, 50 seabass samples were assayed. All the tested 74 SNPs showed clusters plots defined as 'failed', giving no possibility to enlarge the seabass panel with seabream SNPs. When results of the above mentioned 78 SNPs were checked on 23 wild seabream samples, 24 SNPs seemed to work pretty well and 13 SNPs showed lower quality results; all, or at least some of, these SNPs could be potentially added to a future new array in seabream.
The seabream microsatellite multiplex revealed to be highly efficient, with a rate of unambiguous allocations of 100 % for IRTA offspring (99.9 % of single match and 0.1 % of multiple match solved by stochastic approach) and 99.2 % for SME offspring (97 % of single match and 2.2 % of multiple match solved by stochastic approach). The remaining 0.8 % is represented by unsolved multiple assignments.
The seabass microsatellite multiplex gave contrasted results. The rate of unambiguous assignment reached 86.8 % in the IRTA offspring against 76.8 % for Cupimar and 50.3 % for TNSL. Looking at the tank level in the IRTA samples, we note that this rate rises to 95.3 % in V8 against 76.2 % in the V6 tank. The lower rates obtained in seabass can be explained in part by the presence of unsexed breeders in the farm broodstock and mixing of batches of different origins but also by possible genotyping scoring errors.
The 69 SNP panel tested on revealed to be very powerful, with 100 % of unambiguous allocation of which, 99,1 % solved in single match by deterministic approach and 0,9 % solved by stochastic approach (see deliverable 4.17 for more details). The lower rate obtained on seabass (71 %) can be interpreted by:
i) the presence of unsexed breeders in the seabass broodstock and erroneously mixed batches;
ii) the minor seabass DNA quality due to less performing equipment used for the DNA extraction on seabass (PrepStation versus BioSprint on seabream) and
iii) the supposed lower quality of the SNPs reads due to the use of the AutoLoader instead of the AccuFill machine, the latter being used to load the correct quantity of the DNA in the array.
Total number of seabream fish analysed with:
STRs: 4856
1608: SMEs offspring (802 ARDAG and 806 GMF) [4 batches (about 200 samples/batch)]
2960: IRTA offspring [daily and weekly spawning]
197: broodstock (112 ARDAG, 64 GMF, 24 IRTA).
SNPs: 514
226: SMEs offspring (ARDAG and GMF)
288: broodstock (ARDAG, GMF, IRTA, Cupimar, TNSL).
Total number of seabass fish analysed with:
STRs:3435
1634: SMEs offspring (802 ARDAG and 806 GMF) [4 batches (about 200 samples/batch)]
1566: IRTA offspring [daily and weekly spawning]
235: broodstock (116 TNSL, 30 IRTA, 39 CUPIMAR and 50 GMF).
SNPs: 859
481: SMEs offspring (312 Cupimar, 169 TNSL)
143: IRTA offspring [Daily and weekly spawning]
235: broodstock (116 TNSL, 30 IRTA, 39 Cupimar and 50 GMF).
WP4
The aims of WP4 were to:
- develop an allocation software working with STR and SNP markers to allow the accurate and efficient reconstruction of sibship with an easy-to-use procedure,
- perform the sibship reconstruction on every fry sample produced and compare the efficiency of the two types of markers (STRs and SNPs) to increase the rate of correct parentage allocation.
Tasks 4.1. and 4.2. New algorithms to solve multiple matches and produce virtual offspring batches
Associated deliverable:
D4.15: Software with simulation and parenthood procedures
A new version of the Windows-based software for parental allocation (PA) has been developed for the REPROSEL project. The software has been named parental allocation in diploid fish (PAF).
PAF uses the exclusion method as the main/first approach to allocate the parentage. In case of multiple allocations, it uses the likelihood approach. The identification of a parental pair at a given n-locus, with no previous information, is computed according to Marshall et al. (1998). For all the loci, the total likelihood (TL) is computed and, in case of multiple allocations, the difference between the two consecutive ranked TL values (?P) is computed. When ?P > 0 the parental allocation with the highest TL value is accepted and the multiple allocation is resolved. The program follows a stepwise procedure allowing the progressive exclusion of markers, which appears to be useful for SNPs due to their higher global risk of mismatch. Thanks to this procedure PAF distinguishes between the common ±2 pb errors and larger genotypic discrepancies.
The simulation-task, with the creation of Virtual Offspring, allows estimating the efficiency of the loci used in parentage assignment by the finding the ?P threshold that permits to obtain no more than 5 % of erroneous allocations (case of multiple allocations). A mating design is generated based on the parents' dataset. The percentage (or number) of males and/or females involved in the mating can be parameterised. The simulation-procedure produces a virtual offspring by random-sampling all the possible gamete combinations derived from the broodstock. This task can be used only for dataset with gender information on the broodstock.
Two input files are required, the parents and offspring files. They are organised in columns indicating the individual ID, marker ID, biallelic genotypes and sex for the broodfish, and can be sorted in whatever order. The missing values are automatically coded as 9 and SNPs scores (A, C, G, T) are converted in numbers (10, 20, 30, 40). The files are in dbf formats and can be easily generated by a simple copy / paste operation from any tab delimited text file. The call of the markers can be done by simply clicking on a button or pasted as text (markers are separated by a comma).
The main outputs are the parental allocations with their probability estimation, the crossing scheme, the observed and expected heterozygosities and the PIC and exclusion probabilities. They are presented in txt or dbf formats but can also be produced in other formats when requested.
Tasks 4.3. and 4.4. Reconstruction of sibship and cost-effectiveness of the marker panels
Associated deliverables:
D4.16: PA obtained on all fry samples
D4.17: Cost-effectiveness of SNPs and STRs panels for PA.
PA (additional information under WP3):
The PAs were obtained with PAF using the two allocation methods (exclusion and probabilistic). Large genotypic discrepancies were handled accepting a maximum of 2 mismatches for the STRs panels and 3 mismatches for the SNPs panels, while allowing for any ±2 pb errors.
The majority of the offspring was allocated with the STRs panels (13 and 14 markers for seabream and seabass, respectively) while a minor part derived from the SNP validation procedure was obtained with the SNPs panels (69 and 78 markers for seabream and seabass, respectively). Some undetermined broodfish could be sexed and some offspring batches that had been erroneously mixed were evidenced thanks to the allocation results.
Cost-effectiveness:
The allocation efficiency has been tested looking at the effect of increasing the number of breeders (up to 276) to allocate a fixed number of seabream offspring belonging to known families (107 from ARDAG and 119 from GMF). Due to biases in the seabass batches (broodfish incompletely sexed and offspring lots erroneously mixed), the case study has been limited to the seabream batches only. Breeders having more than 15 un-scored SNPs were preventively eliminated (12 in total) in order to avoid biased multiple allocations. This precaution needs to be pointed out for SNPs as the absence of genotype is interpreted by the allocation software as a valid match and thereby increases the percentage of false assignments.
When breeders from the original experimental tanks are considered, the 69 SNPs panel appears to outperform the 13 STRs multiplex by assigning all the true parents in a single match and by solving STRs unassigned multiple matches. Instead, the use of larger (fictive) broodstock increases the rate of erroneous assignments for both types of markers, in particular with SNPs when true and putative parents have similar genotypes. However, when breeders share a limited number of common alleles, SNPs are more efficient than STRs to find the correct parental pairs, allowing in the case of the ARDAG batch a 100 % unambiguous allocation when all the 276 fictive breeders are considered.
The informativeness of the seabass and seabream markers panels were calculated with the polymorphic information content (or PIC). The low PIC values of the SNPs, due to their low allelic diversity, is compensated by non-exclusion probability values similar to STRs for seabream, higher for seabass but still close to zero, indicating a high capacity of both marker panels to exclude false random parents. Removing markers in the panels (two STRs or four SNPs) did not change significantly these values.
When comparing the typing with the two types of markers, we note that STRs often becomes tedious and time consuming, and represents a real bottleneck when high number of samples is processed. In the case of SNPs, the sample processing may be completely automated and the genotype can be obtained rapidly without requiring manual checks. At present, at least four manufacturers propose genotyping solutions adapted to parentage allocation: Illumina, Sequenom, Life Technologies (Applied Biosystems) and Fluidigm. They propose flexible array combinations to choose the genotyping platform adapted to every specific need (how many markers on how many samples).
WP5
The aims of WP5 were to optimise the design of single and multiple trait mass spawning selective breeding schemes taking into account the reproductive features of seabass and seabream populations of the SMEs.
Associated deliverables:
D5.18: Optimised mass spawning breeding scheme designs for commercial broodstocks
D5.19: Guidelines on breeding schemes implementation.
In collabouration with each of the SME partners, information from WPs 2 to 4 on the achieved contributions of males and females in a typical mass spawning tank in their facility for European seabass or gilthead seabream has been used to give an advice on a design of a breeding program. Computer simulations of their specific designs have been done for this optimisation step, taking into account of the reproductive limitations of their respective species. This work was done in close collabouration between the SMEs and the R&D partner Nofima. The result has been delivered in the forms of reports containing information about the selection procedure, the trait to select for, how many tanks needed and other information specific to each of the SMEs. In addition, Nofima has provided the SMEs with a software to select males and females while controlling the rate of inbreeding to the desired rate of 1 %, which can also be used to choose which parents to mate.
A guideline on practical seabass and seabream breeding schemes has been written. The aim of the guideline is to provide the SMEs with general recommendations of the different steps required for setting up a practical mass spawning breeding scheme for seabass and seabream. When differences between species are important for the breeding design, their specific characteristics have been taken into account. Differences in production systems and breeding stock between the SMEs have been considered in the specific SME reports of deliverable 5.18.
The guideline first explains the basis of selective breeding theory, which is an important prerequisite for understanding the guidelines for the mass spawning breeding scheme. Secondly, the guidelines explain how to set up a base population, which should contain high genetic variation, because this population serves as the basis for all coming generations. SMEs are provided with guidelines on how to choose their base population.
Breeding designs are optimised by comparing obtained rate of inbreeding and genetic gain using a stochastic simulation model of the breeding program. In the simulations the range of each parameter on input is from participating SMEs and results obtained in the REPROSEL WPs2 to 4. The focus is on how to obtain highest possible genetic gain per generation when at the same time keeping rate of inbreeding at a low level. In the basic mass spawning breeding scheme, selection of male and female brood fish is done in two steps. There are 50 selected sires and 100 selected dams, which are distributed into 5 mass spawning tanks. These sires and dams are mated randomly, all have different contributions and are producing a total of 10 000 offspring in 5 different mass spawning tanks. After hatching and the grow-out period, 250 males and 250 females are preselected out of the 10 000 fry based purely on their phenotypic weight record. These 500 preselected fish then need to be genotyped in order to assign them to their parents. When the parentage assignment is done the fish are added to the pedigree and BLUP breeding values are calculated based on records of weight of the fish, and relationship within the current generation as well as other generations of fish. From these preselected fish, the best dams and sires with respect to breeding values are selected. These selected sires and dams then go into the mass spawning tanks and become potential parents of the next generation. Due to the different generation intervals, females and males used in each year do not belong to the same year class. This will have a positive effect of the rate of inbreeding per year.
Results show that with respect to rate of inbreeding and genetic gain, 500 to 1000 genotyped offspring seems to be optimal. However, this needs also to be compared with the cost of genotyping. Results also show that based on the contribution of males and females applied in this study, at least 5 different tanks are needed to in order to keep the rate of inbreeding below the general accepted limit of 1 %. A tool to control inbreeding in a practical mass spawning breeding scheme is also presented and has been provided to the SMEs. The tool puts a cost-factor on inbreeding to the estimated breeding values, such that parents are selected based on their relationship in order to get a rate of inbreeding below 1 % per year. In the guidelines it is also discussed how to be able to select for multiple traits. By applying the suggested breeding schemes, it is possible to achieve a genetic gain in growth of at least 10 % per generation.
Potential impact:
In a general way the role played by the genetic markers in selective breeding programs is constantly increasing. In REPROSEL, these markers are first intended to rebuild the pedigree of fish originated from different families and mixed from the egg stage. But such markers need to be cost-effective to change the breeding practices and be more widely used in commercial programs. We believe that the SNP panels developed during the project have the requisites to be effective for parentage inference and also to lower the genotyping costs. The current and near future high throughput genotyping technologies should make the difference and limit the cost to some Euros per individual (today the floor price is inferior to EUR 0.07/SNP).
Cost-effective markers represent as well a strong incentive to implement a breeding program as they permit the rearing tanks used for production to be easily converted for the convenience of the program, so avoiding costly investments for specific facilities and associated high running costs.
The current reproduction protocols suffer from knowledge gaps on mass spawning control. The results of WP2 show how strongly biased the family structure of a traditional offspring batch can be for both studied species. The mass spawning protocols and guidelines (D2.6 and D2.7) explain how to take into account of the biasing factors (maturation stage, sex ratio, number of breeders, tank size, egg collection) and produce the breeding population with optimised parental contribution.
The allocation software PAF has been widely tested on the progeny batches produced during the project (over 8000 offspring from 8 different broodstock) while a part of the assignment results has been compared with other allocation programs (CERVUS and VITASSIGN) to validate the algorithms. The successive versions of the program have finally produced an efficient and easy-to-use tool working both on STR and SNP markers. The scalability of the outputs allows a smart interpretation of the results, starting from the turnkey mating scheme up to key information such as marker incompatibilities and allocation power. In synthesis, PAF is designed for a direct use in the industry and is perfectly adapted to SNPs, presented as valid alternative to STRs.
The guidelines on breeding schemes suggest preselecting a subgroup of the selection candidates of 500 - 2000 fish and, when possible, choose groups of 100 to 200 fish at the time for genotyping. This procedure permits to limit the cost of the program by containing the costly parentage assignment expenses. Having the pedigree information allows calculating the breeding value of the candidates and so getting higher accuracy in selecting the heritable component of the trait under improvement. The breeding scheme describes also a method to keep the rate of inbreeding around 1 % by applying a cost factor parameter when calculating the breeding value. This cost factor can be easily set by using the executable tool provided to the SME partners.
The project results as a whole will contribute to the implementation of sustainable breeding programs. Having the tools to obtain the pedigree information and to increase the effective population size, the SMEs will improve their broodstock while keeping the rate of inbreeding under control. They will increase their competitiveness on the fry market where the share of selected animals is growing continuously.
The foreground on mass spawning will serve the breeding program but will also be useful in production to improve the broodstock management in the hatchery and the fry throughput in the nursery.
Dissemination activities
Communication strategy
The plan to reach the target public has been initiated following the statements of the consortium agreement regarding the confidentiality of the foreground and the publication rights arising from it. The main undertaking was not to jeopardise the immediate and exclusive exploitation of the results by the participating SMEs.
During the course of the project, further discussions showed a divergence between the academic achievement criteria and the industry expectations regarding the dissemination strategy. To help define a common approach, the four participating SMEs were asked to express their preferences on the way each specific foreground should be managed. The questions related to the project foreground were:
I: Do you consider that the information should not be released for now because it may put at risk the exploitation plans?
II: Do you consider that the information can be published outside the consortium now (scientific journals, press release, website public area...)? prior viewing by the SMEs is needed?
III: Do you consider that the information is worth to be licensed to third party upon specific agreement?
IV: Other choice if the above propositions do not suit you?
Any release of non-confidential information outside the consortium was done under the condition of the prior consent of the SME representatives. The release of this information has been done through two main channels:
A. Project website (see http://www.reprosel.eu online)
The REPROSEL website has a public area dedicated to the dissemination of the project as a whole. It gives a general presentation of the project partnership, objectives and main workpackages. The 'Results' section allows reaching a large public with the downloadable abstract of the main project results, while avoiding to give sensitive details to preserve the SMEs interest.
B. Congresses and conferences
To disseminate towards the industry and the scientific community, the REPROSEL results were presented in various European conferences between May and September 2012.
Dissemination actions
Three European conferences have been chosen to present the results of the project:
- 1st Balkan Conference on the Biology of Reproduction in Farm Animals and in Aquaculture, Palace of Culture, Tirana, Albania, 23 - 25 May 2012,
- 2012 Annual Symposium of the Fisheries Society of the British Isles, University of East Anglia, Norwich, UK, 9 - 13 July 2012,
- Aqua 2012 Global aquaculture - Securing our future, Prague, Czech Republic, 1 - 5 September 2012.
In all, four oral presentations and one poster have been presented.
Although not part of the REPROSEL expected results, the work about the spawning behaviour of gilthead seabream derives from the REPROSEL broodfish tanks of IRTA and provides useful information to depict the reproductive behaviour of the species.
The presentations (and their abstracts) can be consulted in the 'Results' webpage (see http://www.reprosel.eu/SitePages/Results.aspx online).
Two other publications referring to the approach explored in REPROSEL have been written in 2012. One is a book chapter to be published in 2013, the other work is already published in a peer-reviewed periodical.
The peer-review publications have been limited not to disclose sensitive information that may jeopardise the future exploitability of the results. However, the project findings will allow in the near future the RTD partners to disseminate more widely the results in scientific journals.
In a more general way internet has also been used to advertise the project, as have done some partners on their website:
Nofima:
http://www.nofima.no/en/prosjekt/reprosel
IRTA:
http://www.irta.cat/en-us/rit/projectes/pages/ProjectDisplayPage.aspx?UrlCode=844
ISILS:
http://www.istitutospallanzani.it/scheda.php?id=133
Exploitation of the results
Ownership of the project results
The foreground of the project is the sole and exclusive property of the SMEs. The joint ownership has been chosen for all the results except for one confidential deliverable (D5.18) specific to each SME.
Presentation of the results
An important step before the delivery of the project results to the SMEs has been the training sessions organised by the RTD partners during the last project meeting. The SME partners took actively part to the sessions, involving also other technical and managerial personnel of the companies. For each of the four workpackages related to RTD activities (WPs 2 to 5), half a day has been dedicated to the presentation of the results followed by an open discussion on the way they should be applied. Demonstrations have been made on how to use the allocation software and how to estimate essential parameters in a breeding programme. The ppt presentations are available in the private REPROSEL webspace at 'WP / Meetings / Final meeting Heraklion'.
A person external to the REPROSEL consortium (Cameron Brown from the Cyprus University of Technology) having done pioneering work on seabream parental contribution, has presented unpublished results of interest for the partners.
As most of the results were not fully completed at the last meeting or needed to be improved further to specific requests, the definitive results have been delivered to the SMEs as a unique package later in December.
How the foreground will be exploited
The advancement level of the genetic program is not the same in all the SMEs, which explains the different approaches and timelines to use the results. Practically, with the project results in hands the SMEs will be able to define their own strategy regarding:
- the base population to start the breeding programme,
- the number of breeders and the sex-ratio forming the breeding nucleus,
- the number and size of the tanks holding the breeders,
- the broodstock rearing protocol (including hormonal treatment) to be used for mass spawning and optimised family structure,
- the size, number and frequency of the progeny batches to be sampled,
- the way each batch should be reared (mixed vs separate) before tagging,
- the selected trait and the recording age / size (sex differentiation, precocious selection),
- the number of fish to be measured, PIT-tagged and fin-clipped,
- the markers panels to be used, the genotyping platform and, associated to it, the genotyping laboratory,
- the way the fish will be genotyped, at once or in a stepwise way,
- the way each candidate will be selected, based on its phenotype or its breeding value (adjusted or not according to their parentage relationship),
- the number of selected males and females and the associated selection intensity,
- the way the new broodfish tanks will be constituted,
- the way each generation will be produced (interval and overlapping).
The guidelines of spawning protocols (D2.6 and D2.7) and breeding schemes (D5.19) indicate for each species how to set up the broodstock groups and collect the eggs to optimise the parental contribution. The screened markers (D3.9 and D3.10) and cost-effectiveness study (D4.17) lists the best performing panels and the current and future genotyping solutions to get the pedigree of the selection candidates. A list of biotech companies providing a genotyping service will help SMEs to choose the best laboratory adapted to their needs. The PAF allocation software (4.15) has been developed for kinship reconstruction with both types of markers; it presents an easy-to-use interface and provides useful outputs for the industry. The cost-factor executable (D5.19) allows the SMEs to calculate the breeding value of the candidates while accounting for the inbreeding level. Finally, a confidential document describes the main steps to implement a breeding scheme adapted to the specificities of each SME (D5.18).
At present no IPR measure has been taken by any of the SMEs, although an interest exists to exploit the marker panels and the allocation software through licensing.
Project website address:
http://www.reprosel.eu/
Contact details:
Nofima:
Herve Chavanne
E-mail: herve.chavanne@nofima.no
Anne Risbrathe
E-mail: anne.risbraathe@nofima.no
Telephone: +47-649-70326
Anna Kristina Sonesson
E-mail: anna.sonesson@nofima.no
Telephone: +47-649-40849
HCMR:
Costas Tsigenopoulos
E-mail: tsigeno@her.hcmr.gr
Telephone: +30-281-0337854
IRTA:
Neil Duncan
E-mail: Neil.Duncan@irta.cat
Telephone: +34-977-745427
ISILS:
Katia Parati
E-mail: katia.parati@istitutospallanzani.it
Telephone: +39-036-378883
GMF
Nikolaos Papaioannou
E-mail: papaioannou@irida-sa.gr
Telephone: +30-697-8182023
ARDAG
Glen Pagelson
E-mail: glen@ardag.co.il
Telephone: +97-286-335111
Cupimar:
Alfonso Vidaurreta
E-mail: campillo@cica.es
Telephone: +34-956-883447
TNSL:
Carlos Mazorra
E-mail: carlosmazorra@tinamenor.es
Telephone: +34-942-718020,
