Introduction
Turbot is the most flatfish species produced in the world, China accounting for more than 65 000 tons in 2015. Selective breeding programs emerged in the early 1990s in Spain and France. Still, few heritability and genetic correlation estimates have been reported regarding edible yields, which are lethal phenotypes to measure. Recently, Schlicht et al (2019) filleted turbots and estimated a heritability of 0.15 for fillet yield for sib-selection. At this time, there is no report of relationship between morphometric and non-lethal traits into a variable that may predict the fillet yield of candidates. The aim of our study was thus to estimate the genetic parameters of non-lethal measurements to evaluate the possibility to predict edible yields on live selection candidates in a French commercial line.
Material and Methods
The fish were derived from the France Turbot Ichtus breeding program in August 2018, using 53 sires and 13 dams mated in a partial factorial design by controlled matings, producing 87 full-sib families. Larvae were reared in common environment. Fishes were individually tagged at 666 days post hatching (dph) (~270g). 897 progenies and their parents were tissue-sampled for DNA parentage assignment using a 96 SNPs panel of markers developed in this project (Gentyane, Clermont-Ferrand, France). The turbots were reared until slaughtering at 750 g (1013 dph) without any selection. The individual body weight (BW) was recorded as well as the weights of different parts: dorsal, ventral and head with abdominal cavity. They were modeled with a linear regression with BW to obtain phenotypic and genetic parameters for residual dorsal part weight (rDW), residual ventral part weight (rVW), residual head and cavity weight (rHCW). The sum of dorsal and ventral parts was used as surrogate for the edible yield relative to BW (rDVW). Non-lethal measurements were made: an individual picture was taken as well as 3D coordinates (Microscribe G2X, Revware Inc., North Carolina) at 19 landmarks on each fish. Geometric morphometrics and computation (MorphoJ and R software) enabled to define new traits such as lengths, areas and volumes. This large number of variables (~50) was evaluated using the Regbest function (in R) to define models that predict the yields. These are the best compromises in terms of model accuracy, percentage of prediction and easy to get variables from field data (e.g. whole-body area, area of the head, distance from snout to operculum, etc.). Heritabilities (h²) and genetic correlations for all traits were computed based on multivariate linear mixed animal models fitted by restricted maximum likelihood in VCE (v6, Groeneveld et al., 2008).
Results and Discussion
The assignment rate of the progeny to their parents using APIS software (Griot et al., 2019) was 78% (the reason for this lower than expected result is still being explored), the dataset comprising then 707 turbots from 8 dams and 29 sires (56 full-sib families). The edible yields (expressed as residuals) had low to intermediate h²: 0.12 ± 0.06 for the dorsal part, 0.16 ± 0.07 for the ventral part, 0.23 ± 0.10 for the head and cavity part, and 0.28 ± 0.10 for the sum of edible parts (Schlicht et al 2019 get a heritability of 0.15 ± 0.03 in the range of what we estimated). Their predictors had higher h² (~0.30 ± 0.11) (Table 1) but did not differ significantly. Phenotypically, the predictors predicted only a limited proportion of the real yields: 8% for the dorsal part, 9% for the ventral part, 15% for the head and cavity, and 15% for the edible parts. However, the genetic correlations between the real yields and their predictors were strong: 0.47 for the dorsal, 0.82 for the edible parts, 0.90 for the head and cavity part, and up to 0.99 for the ventral part.
The predictors defined in turbot in this study explained a lower proportion of the real yields to predict compared to previous studies in round fish such as rainbow trout (37% for carcass yield, Haffray et al., 2013), common carp (63% for headless carcass yield, Prchal et al., 2018), European sea bass (27% to 38%, Vandeputte et al., 2017). However, the genetic correlations were much higher in our study. And because the predictors are heritable traits, the selection on yield predictors in turbot may be at least as effective as relying on family data acquired by slaughtering the sibs to know the real yields.
Conclusion
Predictors are moderately heritable and positively genetically correlated with the real yield they predict. This means that indirect predictors can be used to index candidates themselves through the use of non-invasive measurements that should enhance accuracy of breeding evaluation, a work in progress.
Acknowledgement
The data presented here were obtained in the TURBOOST (2018-2022) project which received funding from the European Maritime and Fisheries Fund (FEAMP) and by the French Government through the FranceAgriMer national body.
References
Griot, R., et al., 2019. APIS: An auto-adaptive parentage inference software that tolerates missing parents. Mol. Ecol. Resour. 00:1-12
Groeneveld, E., et al., 2008. VCE User’s Guide and Reference Manual Version 6.0. Friedrich Loeffler Institute, Neustadt, Germany, 125 pp.
Haffray, P., et al., 2013. 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. 410-411, 236-244.
Prchal, M., et al., 2018. Potential for genetic improvement of the main slaughter yields in common carp with in vivo morphological predictors. Front. Genet. 9:283.
Schlicht, K., et al., 2019. Estimation of genetic parameters for growth and carcass traits in turbot (Scophthalmus maximus). Arch. Anim. Breed. 62:265-273
Vandeputte, M., et al., 2017. Investigation of morphological predictors of fillet and carcass yield in European sea bass (Dicentrarchus labrax) for application in selective breeding. Aquaculture.