A review of the use of genetic markers applied to aquaculture and wild fish populations was conducted. The review detailed current knowledge and provided a foundation on which to explore new opportunities to improve the utility of genetic markers in fish populations. The current use of markers is diverse and includes but is not limited to: improvements to genetic selection of captive populations (marker assisted selection, genome wide selection, walkback selection), estimation of variance components (heritability, genetic correlations), pedigree identification (relatedness, kinship, inbreeding, genetic tagging), population studies (stock differentiation, migration rates, species identification, invasive species distribution, effective population size, phylogenetics, illegal fishing) and survival estimation.
This research was used to build on the utility of existing applications of genetic markers in fish populations. In the first application genetic markers were used to identify sires of progeny in a novel breeding program using in vitro fertilisation of eggs. Computer simulation between family selection at the onset of a breeding program was optimised to yield a 40% increase in growth rate for barramundi (Lates calcarifer). Such a large gain is of significant economic importance to the barramundi aquaculture industry. The breeding program was designed to yield additional genetic gains from long term selection by managing inbreeding.
Following on from this related design a new method using a binary threshold model was developed to rapidly assess genotype by environmental interactions (GxE) by modelling and estimating genetic correlations between environments. The motivation for this study was to improve the ability to estimate how much genetic improvement predicted by a selective breeding program will be realised in the commercial environment. As such, rapid estimates of genetic correlations are important during the very early stages of breeding program investment. The design was suitable for rapid assessment of GxE over one generation with a true 0.70 genetic correlation yielding standard errors as low as 0.07.
While the first two applications assumed sufficient genetic markers to identify sires, an accurate new statistical method was developed to identify individuals using genetic markers when Type I and Type II errors occur. The new theory advances the application of likelihood methods which were implemented in a new software tool called SHAZA. The methods are useful in wildlife forensic studies where missing data can often occur as a result of the breakdown of DNA in field conditions.
The theory developed and implemented in SHAZA was applied to a mark-recapture design in a dataset which had many individual genotypes with missing loci. The new theory was able to recover 80% more pairwise comparisons of individual genotypes compared to discarding missing data. This facilitated the estimation of abundance and proportion of Spanish mackerel (Scomberomorus commerson) caught during commercial harvest which was determined with finite estimates. The abundance estimate is potentially useful in fisheries management and ecological monitoring.
The utility of genetic markers to assess an idealised estimate of the abundance of breeding adults was also investigated by estimating effective population size (Ne). In this study it was discovered that outlier genotypes on non-conspecific species created a large bias in linkage disequilibrium estimation of Ne. Correspondence analysis methods were tested using simulation as a means of identifying non-target species. Simulations showed that the identification and removal of these non-target genotypes was successful in improving the accuracy of Ne estimation.
A review of the major findings, their implications, caveats and future research ideas are discussed in the final chapter. One future area of research would be to investigate the utility of estimating the percentage of full sibs in populations that have insufficient genotype data collected for individual assignment tests. A new proposed method is based on a curve of the cumulative number of false positives plotted against the log likelihood ratio of pairwise comparisons. The change in the curve was found to be sensitive to small changes in the percentage of full-sibs in a population.