ACCURATE GENOMIC PREDICTION OF PHENOTYPIC SEX WITHIN EUROPEAN SEA BASS FAMILIES: A UNIQUE TOOL TO UNDERSTAND AND CONTROL SEX RATIO VARIATIONS IN A SPECIES WITH POLYGENIC SEX DETERMINATION

The European sea bass, Dicentrarchus labrax, a major aquaculture species in Europe, has a peculiar polygenic system of sex determination, with several QTLs determining the genetic component of sex. Temperature during the early phases of larval rearing (before 60 days post hatch, dph) also has large effects on sex-ratio: high temperatures (>17°C, HT) masculinize fish that would have developed as females in low temperature (<17°C, LT). In a polygenic system, there cannot be a diagnostic sex marker, and thus it is a priori impossible to identify which individuals are sex-reversed by temperature (i.e., the genotypic females that differentiate as males, called "neomales" in fish with chromosomal sex determination).

We produced two families by mating two males with the same female. The progenies were reared in common garden, half of them in a sex-neutral LT temperature protocol (16°C until 60 dph), and the other half in a masculinizing HT treatment (21°C until 60 dph). The fish were sexed at 319 dph, and assigned to their parents with 12 microsatellite markers. Among the assigned fish, 92 HT fish and 84 LT fish from sire A, as well as 78 LT fish from sire B, were selected for genotyping on a 3K SNP Illumina ISelect array. In total, 966 SNP markers were polymorphic within or between the two families, with a call rate >0.90, a sufficient minor allele frequency (MAF>0.05) and no mendelian transmission errors.

Sex was predicted by the genomic estimated breeding value of sex tendency using a genomic best linear unbiased prediction (GBLUP) model, applied to each family at each temperature and various combinations thereof. When data from two temperature treatments were applied, a fixed effect of temperature was included in the model. To the prediction of sex, a receiver operating characteristic (ROC) curve was built using all possible sex tendency thresholds to separate males from females. The quality of the prediction was assessed by the area under the curve (AUC).

The AUC was close to its maximum value of 1 in family A, HT (AUC=0.999). In family A, LT, the AUC was 0.947, and it was 0.994 in family B, LT. Thus, prediction of sex was very good in the three groups. When the data from both LT and HT were combined in family A, the AUC was 0.962, showing coherence between the models at two temperatures. However, when data from family A and family B were combined, the AUC dropped to 0.900, showing that the optimal model was not the same in both families.

We show that despite its polygenic nature, sex in The European sea bass can be predicted with very high accuracy within family using a genome-wide approach (GBLUP). This paves the way to the identification of neomales and to the possibility to identify the phenotypic sex of individual animals at the time of sex determination, which should provide major advances to our understanding of European sea bass sex determination.

This study was supported by the EU's 7th Framework Programme under grant agreement 262336 (AQUAEXCEL), TNA project 0102/06/07/20, and by the French Ministry of Environment under grant CRECHE²⁰¹⁶