The effects of penalization of FAMACHA© scores of lambs treated for internal parasites on the estimation of genetic parameters and prediction of breeding values

Internal parasite infestation in sheep can be practically assessed using a subjective score for the color of the inner eyelid (FAMACHA© score). Additive genetic variation for this trait would permit selection against internal parasite infection or for tolerance to internal parasites. However, medicated sheep have improved FAMACHA© scores shortly after treatment; use of these records could produce misleading breeding values. The objectives of this study were to (1) apply different penalties to FAMACHA© records of lambs that required treatment, and thereby generate distinct data sets associated with those penalties, and (2) estimate heritability and predicted breeding values for each data set under moderate and severe internal parasite challenge conditions. Two types of penalties were applied. The first penalty consisted of adding the average increase (worsening) in FAMACHA© score of untreated lambs to the records of the treated lambs in a given year. The second type of penalty was the assignment of a high value as the actual FAMACHA© score of treated lambs; in multiple analyses this value was 5 (which is the maximum possible FAMACHA© score), or increased to 6, 7, 8, 9, or 10. Single-trait animal models were employed in analyses of FAMACHA© scores in Merino lambs (n=1671) across 6 yr. Results indicated that, under both moderate and severe internal parasite challenge conditions, application of penalties increased the additive genetic variance (and therefore the estimated heritability) relative to analyses in which records of treated lambs were included without any modification or analyses in which they were excluded. Estimates of correlation coefficients of predicted breeding values for FAMACHA© score from the various analyses indicated that, in most cases, there were strong positive associations among those from the penalized data sets. Correlations of ranks of predicted breeding values were also strongly positive; however, those involving penalty values of 10 were lower, indicating some rank changes. Distributions of predicted breeding values indicated that in moderate internal parasite challenge conditions, the analyses of penalized data removed multi-modality, but resulted in highly leptokurtic distributions (many observations congregated at the mean). In the severe internal parasite challenge conditions, applications of penalties induced multi-modality. Penalizing records of treated lambs with a value of 5 appears to be a reasonable method for more closely estimating the additive genetic variance and therefore, the heritability, which may permit more useful genetic analyses. Excessive penalization of records of treated lambs appeared to be unnecessary at best, and at worst detrimental to genetic evaluation through rank changes of predicted breeding values.