Genomic selection for production of tilapia (Oreochromis niloticus) resistant to Francisella noatunensis

• Grant number: 23/01942-1
• Support Opportunities: Scholarships in Brazil - Post-Doctoral
• Start date: April 01, 2023
• End date: March 31, 2026
• Field of knowledge: Agronomical Sciences - Fishery Resources and Fishery Engineering - Aquaculture
• Principal Investigator: Diogo Teruo Hashimoto
• Grantee: Baltasar Fernandes Garcia Neto
• Host Institution: Centro de Aquicultura (CAUNESP). Universidade Estadual Paulista (UNESP). Campus de Jaboticabal. Jaboticabal , SP, Brazil
• Associated research grant: 21/11955-8 - Solutions for emerging diseases in fish farming: diagnosis, vaccines and breeding, AP.CCD

Abstract

Nile tilapia (Oreochromis niloticus) is one of the most important farmed freshwater fish in the world. The four main Nile tilapia producer countries are currently China, Indonesia, Egypt and Brazil, representing together more than 83% of world production in 2021. One of the main challenges for the advancement of tilapia farming in Brazil is the frequent appearance of diseases during the productive. Among them, Francisella noatunensis subsp. orientalis (FNO). Francisellosis is caused by a gram-negative, facultative intracellular pathogen and causes high rates of mortality and morbidity, especially in the younger stages. In addition, francisellosis is associated with the presence of granulomas in various internal organs, mainly the spleen. Several forms of control have been adopted to combat this disease, for example, the use of antibiotics and vaccines. However, these measures have shown limited efficiency. In this sense, the adoption of genetic improvement programs can represent a viable alternative to control the disease, since it could increase the resistance of the animals against the infection. The aim of this project is to generate a genomic selection program to improve the tilapia production chain, resistant to FNO. The fingerlings used to form the breeding stock will come from different farmers and also from the CAUNESP. A total of 150 families will be formed in individual hapas in the ratio of 1 male to 2 females. After collecting eggs from each family, the animals will be incubated separately under controlled environmental conditions. After hatching, approximately 150 larvae per family will be transferred to hatchery tanks until the fingerling/juvenile stage. Juveniles will be individually identified using microchips (pit-tags). Finally, the animals will be transferred to earthen ponds. About 20 animals from each family will be challenged against FNO at CAUNESP (Jaboticabal, SP) and the phenotypes of binary survival (SB) and time of death (TDM) will be recorded. DNA samples will be taken from all animals challenged for the genotyping step. Genotyping will be done with two different densities: 60K SNPs for the parents and 1K SNPs for the offspring (challenged animals). After genotyping, the genomic data will pass by a quality control process to remove uninformative and low segregation rate SNPs. The genomic data will be used for kinship reconstruction (pedigree formation). With the pedigree and genomic data of low and high density, it will be possible to impute genotypes (predict the missing SNPs in the challenged animals), i.e., we will use the 60K SNPs as a reference to impute the genotypes of the animals genotyped at 1K. This is a cost-effective strategy for obtaining genotypes at higher density. Genome-wide association studies (GWAS) will be implemented using the single-step GBLUP and weighted single-step GBLUP method, combining information from genotyped and phenotyped animals into the H matrix. Through GWAS, it will be possible to unravel the genetic architecture of resistance to FNO searching for regions of the genome that may be associated to this trait. A final genomic prediction study will be implemented to evaluate the accuracy of genomic prediction in comparison to the use of pedigree only. For this, the breeding values (EBVs) of animals will be estimated using the PBLUP model that incorporates only the pedigree information. The genomic breeding value (GEBVs) of animals will be estimated using the GBLUP model. For comparative purposes, a five-folded cross-validation scheme will be implemented by masking part of animals' phenotypes (20%) and using the remaining 80% as a reference. The benefit of incorporating genomic information will be evaluated by comparing the accuracy of predictions using both models with different sources of information.