AQUA 2024

August 26 - 30, 2024

Copenhagen, Denmark

ACCOUNTING FOR A MIXTURE OF DIPLOID AND TRIPLOID FISH IN GENETIC EVALUATION

Julien Roche, Pierre Patrice , Alexandre Desgranges , Pierrick Haffray, Florence  Phocas *

 

* Université Paris-Saclay, INRAE, AgroParisTech, GABI, 78350 Jouy-en-Josas, France

florence.phocas@inrae.fr

 



 Induced triploidy is used in various fish and shellfish species to  sterilize and  improve growth and/or flesh quality in commercial productions, as well as to limit  the risk of genetic introgression of farmed genotypes into wild populations. In particular, rainbow trout production of large fillets is based on triploid fish in fresh water. S elective breeding programs  on sib performances are exclusively performed on diploid information. Maximizing genetic gain requires the evaluation of breeding values of the diploid candidates based on triploid sibs’ performances . In mixed-family designs, this implies genotyping triploids and recovering their pedigree. Until recently,  no reliable and open source tools were available to do so . W e  thus  developed methods  and R packages for genotype calling (https://cran.r-project.org/web/packages/APIS/index.html) to properly answer these issues. T he  relationships  between progeny and parents are  affected by the triploid nature of offspring as triploid fish inherit two copies of DNA from the mother and one from the father.  Therefore, the genetic evaluation gathering information from both diploids and triploids must  also account for the various allelic dosages in the relationship matrix used to run a BLUP animal evaluation. This issue was solved in the current study in order to estimate genetic parameters.

Two sib batches  (one diploid, one triploid)  of  ~1,500 fish each  were  genotyped and phenotyped for production traits at 18 months. They came from the same parental cohort (190 dams  × 98 sires)  produced by Milin Nevez (Bretagne Truite, France). T he BLUPF90 software was used  for the genetic evaluation based on a bivariate animal model integrating the inverse relationship matrix provided by the R package ‘polyAinv’ (Hamilton & Kerr, 2019) that allows various ploidy levels .  Heritability of diploid and triploid performances as well as their correlations were calculated using an AIREML algorithm and either a  full diploid genetic model (wrong model) or a correct model accounting for the right ploidy level of any offspring.  While the  heritability values derived for diploid performances were similar for the two models (see Table below), the  heritability estimates  were largely overestimated for triploids under the wrong model . Using the correct  genetic  model,  heritability estimates were very cl ose  between  diploid and triploid performances. Genetic correlations were less impacted by the genetic model used, and were high in any case (rg> 0.70).

Given the se high correlations  estimated  between diploid and triploid performances , selection based on diploid information is efficient to improve a commercial production with triploid fish.