GENETIC PARAMETERS FOR COMMERCIAL TRAITS IN NILE TILAPIA USING MULTIVARIATE GENOMIC MODELS

Nile tilapia (Orechromis niloticus) is the second most important aquaculture species in the world, supported by commercial production in more than 120 countries. Most of the commercially farmed tilapia originate from the GIFT strain of tilapia, formed by crossing of 4 African wild strains and 4 farmed Asian strains back in 1990s. GenoMar Supreme Tilapia (GST ®) originates from this GIFT population and has continued the breeding program commercially and is currently in generation 28. Genomic selection has shown to increase the predictive ability and genetic gain, across all the species, including the aquaculture. With the availability of the genomic resources for Nile tilapia and the results showing increase in predictive ability and unbiased estimates of breeding values, genomic era for Nile tilapia is inevitable.

The experimental population was obtained from the generation 26 GST® strain. Animals were reared in 8 different batches in the GenoMar Breeding Centre in the Philippines and was harvested after 11 months in the ponds using standard commercial practices. The phenotypes were available for five important traits: body weight at harvest (BW), fillet weight (FW), fillet yield (FY), trash belly (TB) and trash head (TH). DNA was extracted from the fin clips and was genotyped using Onil50 Affymetrix array. The raw dataset contained 58,466 SNPs. After quality control (selection of high quality "PolyHighResolution" and "NoMinorHom" SNPs, maf filtration <0.05, individual call rate <0.9), 48,960 SNPs (83.74%) was available for 2549 animals. Genomic relationship matrix was constructed using Gmatrix software using VanRaden method and the multivariate GBLUP models (with batch, difference of age during harvesting, filleter for the traits FW, FY, TB and TH as the fixed effects; and genetic effect as the random effect) and were run in DMUv6, assuming genetic and residual covariances between the five traits.

These are the first estimates using multivariate genomic models in Nile tilapia. All the five traits were found to have moderate heritability ranging from 0.09 ± 0.02 for FW to 0.21 ± 0.03 for TH (Table 1). FY was found to be negatively correlated with all the traits, whereas the correlation between other traits were very high, meaning selection for one trait will inevitably lead to selection for other traits because same genetic mechanism may be controlling these traits. The future strategy in GST® Nile tilapia will be to find the QTLs for these traits and utilise it in genomic selection. Besides these universal traits, genomics will help in efficient selection for different disease resistance traits.