Short communicationAccuracy of BLUP breeding values in a factorial mating design with mixed families and marker-based parentage assignment in rainbow trout Oncorhynchus mykiss
Introduction
Less than 10% of world aquaculture production benefits from genetically improved seed. When it does, this comes mostly from conventional family-based breeding programs (Gjedrem et al., 2012), using families reared separately until tagging and Best Linear Unbiased Prediction, or BLUP (Henderson, 1984). One of the major reasons for this limited development is the high cost of these programs, which require both initial investment in numerous family tanks (>100) and highly trained staff using very good traceability and rearing practices to limit both pedigree errors and initial tank effect. This is a serious obstacle for small and medium enterprises (SMEs) and for the initiation of selective breeding in new aquaculture species.
Alternative breeding schemes using marker-based parentage assignment and combining mass selection on traits measurable in candidates (such as growth) and sib selection on lethal traits (quality, disease resistance, etc.) have not been greatly studied so far (Vandeputte and Haffray, 2014). This kind of breeding scheme limits both tank effect and initial investment, as all families can be reared together in just one tank. The use of common garden rearing and molecular pedigrees also relieves constraints on family number, allowing the production of more families for a fixed number of parents (e.g., by the use of factorial mating designs). This increases the potential for between-family selection intensity at a given level of inbreeding (Dupont-Nivet et al., 2006; Sorensen et al., 2005).
The relative efficiency of breeding programs can be assessed by the accuracy (rAÂ) of the estimated breeding values (EBVs). This accuracy is the correlation between the true breeding value (TBV) and the EBV, and can be derived from mixed-model equations (Van Van Vleck, 1993).
In aquaculture, there are few reports on the accuracy of EBVs for production traits, with only two references available on conventional breeding schemes with initially separated families (Gonzalez-Pena et al., 2016; Kause et al., 2007) and one on a mixed families scheme (Blonk et al., 2010). One problem with the mixed-families design, using a posteriori parentage assignment, is that the number of sibs per family is variable and unpredictable. In addition, as factorial designs generate very high numbers of families, they can result in very low numbers of fish recorded from each family. Small and variable family sizes could be seen as unfavourable to the accuracy of breeding values.
The aim of the present work was to estimate the accuracy of EBVs for different traits (growth, carcass traits and fillet quality) in a factorial mating design in rainbow trout Oncorhynchus mykiss using a posteriori parentage assignment. Genetic parameters, EBVs and rAÂ were estimated within a single generation using performances collected from a random sample of 2000 offspring. Both univariate and multivariate models were investigated as the literature mostly reports evaluations or simulations with univariate models, but breeding programs generally rely on multivariate models. The accuracy of sib selection was compared with combined family selection for each trait and with univariate or multivariate models using a leave-one-out (LOO) procedure, where the phenotype of any given individual was either used (combined selection) or not (sib selection).
These estimates of accuracy in a mixed-family design in aquaculture are expected to support the introduction of combined family or sib-based selection on quality traits in mass selection programs initially designed to improve growth.
Section snippets
Management and phenotyping of animals
The measured traits and rearing environments were indexed according to the ATOL (Animal Trait Ontology for Livestock) and EOL (Environment Ontology for Livestock) ontologies available on the ATOL (http://www.atol-ontology.com/index.php/en/les-ontologies-en/visualisation-en) website.
The offspring sample had already been used to estimate genetic parameters and design predictors for quality traits in previous studies (Haffray et al., 2012a; Haffray et al., 2013) and was established from a
Results
Basic statistics for all traits measured at slaughter are given in Table 1.
DNA parentage assignment success was 99.5%. All parents were represented in the progeny. The observed number of full-sib families was 559 (93.1% of the expected 600). The average full-sib family size was 3.5 ± 2.2 and varied from 0 to 13. Paternal (or maternal) half-sib family size varied from 8 to 45 (from 12 to 61).
Heritability at processing for the traits previously published (BW, Carc%, Fil%) for this experiment (
Discussion
We estimated genetic parameters and breeding accuracies for individual and sib evaluated traits (growth and quality traits) in a mixed-family factorial mating design. Intermediate to high accuracy of EBVs was estimated even with a low number of full-sibs per family (3.3 on average) for combined family selection and sib-based selection.
Heritability estimates for BW, Fil% and Carc% (already published in Haffray et al., 2012a) and for new traits (Fat% and a*) were in the upper range of previous
Conclusions and perspectives for application
This study is the first of its kind estimating genetic parameters, EBVs and rAÂ of growth, processing traits and fillet quality in a mixed-family rearing design. The accuracy of EBVs was high, at least equivalent with those simulated in conventional (using separate families) breeding programs, and superior to those estimated from conventional breeding program data. This provides new insights for starting selection programs combining mass selection for growth and sib-based selection on quality
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
PH managed the project, estimated the heritabilities and rAÂ and wrote the manuscript, FE estimated accuracy of EBVs and set up the LOO procedure, JB managed the data collection on a* colour and Fat%, HC supervised the estimation of genetic parameters and of EBV accuracy, MDN and MV supervised the genetic data treatment and the writing of the manuscript. All authors contributed to designing the study, interpreting the results, revising the manuscript and read and approved the final manuscript.
Acknowledgements
The authors are indebted to Cyrilles Deshayes and Marine Levadoux from the French Interprofessional Committee for Aquaculture Products (CIPA) for their support and management of the project and the staff of INRA, IFREMER and French breeding companies who are members of SYSAAF (Aqualande, Viviers de France, Font Rome Pisciculture and Les Fils de Charles Murgat) for their participation in the fish rearing and data collection. This work was supported by the French National Interprofessional Office
References (24)
- et al.
Heritability of shape in common sole, Solea solea, estimated from image analysis data
Aquaculture
(2010) - et al.
Effect of different mating designs on inbreeding, genetic variance and response to selection when applying individual selection in fish breeding programs
Aquaculture
(2006) - et al.
The importance of selective breeding in aquaculture to meet future demands for animal protein: a review
Aquaculture
(2012) - et al.
Negative genetic correlations between production traits and head or bony tissues in large all-female rainbow trout (Oncorhynchus mykiss)
Aquaculture
(2012) - et al.
Minimizing maternal effect in salmonid families mixed since eyed stages and a posteriori DNA-pedigreed
Livest. Sci.
(2012) - et al.
Genetic parameters of in-vivo prediction of carcass, head and fillet yields by internal ultrasound and 2D external imagery in large rainbow trout (Oncorhynchus mykiss)
Aquaculture
(2013) - et al.
A survey of cross-validation procedures for model selection
Statist. Surv.
(2010) - et al.
Selection for an improved disease resistance using factorial mating designs and molecular based pedigrees in fish: a simulation study
Official Recommendations on Uniform Colour Space, Colour Difference Equations and Metric Colour Terms
(1976)- et al.
Determination of the lipid content of rainbow trout (Oncorhynchus mykiss) filets with the Torry fish fat meter (R)
Sci. Aliment.
(1998)
Introduction to Quantitative Genetics
Genome-wide association study for identifying loci that affect fillet yield, carcass, and body weight traits in rainbow trout (Oncorhynchus mykiss)
Front. Genet.
Cited by (9)
Realised genetic gains on growth, survival, feed conversion ratio and quality traits after ten generations of multi-trait selection in rainbow trout Oncorhynchus mykiss, fed a standard diet or a “future” fish-free and soy-free diet
2022, Aquaculture ReportsCitation Excerpt :The main advantage of using the control line approach is that selection response can also be estimated for traits which are not recorded in the breeding programme (in our case head yield, vertebral axis yield, fillet yield, body composition – including fatty acids), or even traits that cannot be recorded on an individual basis (FCR, nutrient gain), which have a key importance for production sustainability of the. It has to be noted that although pedigree selection was used in the present breeding programme, it was combined with individual pre-selection for growth and externally recorded quality traits (muscle fat-meter value, morphological index, ultrasound estimates of gutted yield) (Haffray et al., 2018) for which the performance and pedigree of culled animals was not known, thus precluding a straightforward use of genetic trends to evaluate selection gains. Finally, comparing a selected and a control line also enabled us to evaluate their relative performance both at the same age (374 dph) and at the same body weight (1650 g).
Genetic parameter estimates for intermuscular bone in blunt snout bream (Megalobrama amblycephala) based on a microsatellite-based pedigree
2019, AquacultureCitation Excerpt :The number of progeny in 213 full-sib families ranged from 1 to 20 (Supplementary Table S1). Considering the factorial mating design in our study, the offspring data from these full-sib families were all used for heritability estimation as indicated by Haffray et al. (2018). The average counts of IBs was 118 for the whole body, 81 for the dorsal section, 37 for the ventral section, and 59 for left and right sections, respectively (Table 3).
Insights from the early generations of the Swedish rainbow trout (Oncorhynchus mykiss) breeding program
2023, Acta Agriculturae Scandinavica A: Animal Sciences