The influence of infection/expression level on selection response for worm resistance in sheep

Internal parasites cost sheep producers across the world hundreds of millions of dollars each year in treatment and lost production. Genetic variation in resistance to internal parasites exists and genetic progress has been achieved through selection.With the changing climate and sheep breeders adopting genetic tools for resistance to internal parasites the ability to obtain adequate expression is a challenge in some circumstances. Furthermore, the genetic evaluation of worm egg count (WEC) is problematic due to the variable levels of expression across different environments and years, type and representation of various species of internal parasite and the skewed distribution of the data. The skewed distribution of raw WEC observations is a result of the majority of animals having low faecal egg counts and a small number having very high counts. These data are commonly transformed with a cube root in an attempt to normalise the data prior to genetic evaluation. There is a need to explore the potential use of data with lower levels of expression in the evaluation to optimise the use of data. The aim of this study was to examine the impact of expression level on the ability to detect genetic variation and derive more adequate thresholds for the inclusion of data into the genetic evaluation.Data on the performance and pedigree of Merino sheep were obtained from the MERINOSELECT database maintained by Sheep Genetics. All worm egg count records were selected from weaning through to hogget ages resulting in 542,683 records from 503,619 animals.Means, standard deviations and the proportion of observations with a value of 0 were calculated for each CG to use as variables to filter the data for analysis. These three variables were used to filter data with low expression. The three variables were only moderately related (r< 0.55) to each other but all were related to expression of genetic variation. The proportion of observations with a value of 0 was a less reliable predictor of heritability. The results suggest that a combination of CG mean and standard deviation are required. Filtering the data to only use CGs with a mean WEC of at least 4 and standard deviation of at least 1 (on the cube root scale) maintained the heritability above 0.20 for all traits and enabled 91% of the records available to be utilized.The results clearly demonstrated that the amount of variation in WEC influence the heritability of the trait and thus response to selection. These results will inform more appropriate filters to be applied to the routine evaluation for worm egg count in Australia.