The giant freshwater prawn (Macrobrachium rosenbergii) is an economically important farmed species in China, and the genetic improvement of its growth traits is of great significance for enhancing industrial profitability.
Objective This study aimed to investigate the effects of single-step genomic best linear unbiased prediction (ssGBLUP) and best linear unbiased prediction (BLUP) under single-trait and two-trait models on the estimation of genetic parameters and the accuracy of breeding value prediction for body weight and body length in Macrobrachium rosenbergii, thereby providing a theoretical basis for breeding practices.
Method The nucleus population of M. rosenbergii after five generations of selection was used as the research subject. A genetic background analysis was conducted using 278 469 single nucleotide polymorphism (SNP) loci from 316 genotyped individuals across two generations (G04 and G05). Combining the SNP data of genotyped individuals from both generations with the phenotypic records of growth traits from 4 771 individuals in the G05 generation, single-trait and two-trait models were constructed based on ssGBLUP and BLUP methods, respectively. Genetic parameters for body weight and body length were estimated, and the accuracy of breeding value prediction was evaluated.
Result The effective population sizes for the two generations were 138 and 124, respectively, with highly consistent genetic backgrounds, although a certain degree of inbreeding was observed in both generations (genomic inbreeding coefficients ranging from 0.074 to 0.096). The inconsistency rate between sibship identification and the two-generation pedigree was 10.44%. Under both models, the heritability estimates for body weight (0.194±0.086–0.203±0.090 vs. 0.161±0.079–0.172±0.083) and body length (0.195±0.091–0.206±0.091 vs. 0.159±0.081–0.169±0.083) obtained by ssGBLUP and BLUP did not differ significantly. The genetic correlation between body weight and body length was extremely strong (> 0.98). The accuracy of breeding value prediction under the two-trait model (body weight: 0.762–0.769; body length: 0.760–0.769) was significantly higher than that under the single-trait model (body weight: 0.463–0.466; body length: 0.459–0.464), whereas no significant difference in accuracy was detected between the ssGBLUP and BLUP methods.
Conclusion Compared with the single-trait model, the two-trait model significantly improved the accuracy of breeding value prediction for growth traits, indicating that utilizing information sharing between traits can effectively enhance selection efficiency. However, ssGBLUP did not significantly improve prediction accuracy, likely due to the lack of phenotypic records for genotyped individuals and the limited sample size. Therefore, it is recommended that multi-trait models be prioritized in current breeding programs, and future efforts should focus on expanding the population size with both genotypes and phenotypes to fully realize the potential of genomic selection.
DownLoad: